correct rtl_op_set_tim and rtl_update_beacon implementation missing #54

jpbruyere · 2013-10-31T14:14:42Z

patch found at 'http://www.spinics.net/lists/linux-wireless/msg96128.html'

As the new x86 CPU bootup printout format code maintainer, I am taking immediate action to improve and clean (and thus indulge my OCD) the reporting of the cores when coming up online. Fix padding to a right-hand alignment, cleanup code and bind reporting width to the max number of supported CPUs on the system, like this: [ 0.074509] smpboot: Booting Node 0, Processors: #1 #2 #3 #4 #5 #6 #7 OK [ 0.644008] smpboot: Booting Node 1, Processors: #8 #9 #10 #11 #12 #13 #14 #15 OK [ 1.245006] smpboot: Booting Node 2, Processors: #16 #17 #18 #19 #20 #21 #22 #23 OK [ 1.864005] smpboot: Booting Node 3, Processors: #24 #25 #26 #27 #28 #29 #30 #31 OK [ 2.489005] smpboot: Booting Node 4, Processors: #32 #33 #34 #35 #36 #37 #38 #39 OK [ 3.093005] smpboot: Booting Node 5, Processors: #40 #41 #42 #43 #44 #45 #46 #47 OK [ 3.698005] smpboot: Booting Node 6, Processors: #48 #49 #50 #51 #52 #53 #54 #55 OK [ 4.304005] smpboot: Booting Node 7, Processors: #56 #57 #58 #59 #60 #61 #62 #63 OK [ 4.961413] Brought up 64 CPUs and this: [ 0.072367] smpboot: Booting Node 0, Processors: #1 #2 #3 #4 #5 #6 #7 OK [ 0.686329] Brought up 8 CPUs Signed-off-by: Borislav Petkov <[email protected]> Cc: Libin <[email protected]> Cc: [email protected] Cc: [email protected] Cc: [email protected] Cc: [email protected] Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Ingo Molnar <[email protected]>

Turn it into (for example): [ 0.073380] x86: Booting SMP configuration: [ 0.074005] .... node #0, CPUs: #1 #2 #3 #4 #5 #6 #7 [ 0.603005] .... node #1, CPUs: #8 #9 #10 #11 #12 #13 #14 #15 [ 1.200005] .... node #2, CPUs: #16 #17 #18 #19 #20 #21 #22 #23 [ 1.796005] .... node #3, CPUs: #24 #25 #26 #27 #28 #29 #30 #31 [ 2.393005] .... node #4, CPUs: #32 #33 #34 #35 #36 #37 #38 #39 [ 2.996005] .... node #5, CPUs: #40 #41 #42 #43 #44 #45 #46 #47 [ 3.600005] .... node #6, CPUs: #48 #49 #50 #51 #52 #53 #54 #55 [ 4.202005] .... node #7, CPUs: #56 #57 #58 #59 #60 #61 #62 #63 [ 4.811005] .... node #8, CPUs: #64 #65 #66 #67 #68 #69 #70 #71 [ 5.421006] .... node #9, CPUs: #72 #73 #74 #75 #76 #77 #78 #79 [ 6.032005] .... node #10, CPUs: #80 #81 #82 #83 #84 #85 #86 #87 [ 6.648006] .... node #11, CPUs: #88 #89 #90 #91 #92 #93 #94 #95 [ 7.262005] .... node #12, CPUs: #96 #97 #98 #99 #100 #101 #102 #103 [ 7.865005] .... node #13, CPUs: #104 #105 #106 #107 #108 #109 #110 #111 [ 8.466005] .... node #14, CPUs: #112 #113 #114 #115 #116 #117 #118 #119 [ 9.073006] .... node #15, CPUs: #120 #121 #122 #123 #124 #125 #126 #127 [ 9.679901] x86: Booted up 16 nodes, 128 CPUs and drop useless elements. Change num_digits() to hpa's division-avoiding, cell-phone-typed version which he went at great lengths and pains to submit on a Saturday evening. Signed-off-by: Borislav Petkov <[email protected]> Cc: [email protected] Cc: [email protected] Cc: [email protected] Cc: [email protected] Cc: [email protected] Cc: Linus Torvalds <[email protected]> Cc: Andrew Morton <[email protected]> Cc: Peter Zijlstra <[email protected]> Cc: Thomas Gleixner <[email protected]> Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Ingo Molnar <[email protected]>

also fix spellos WARNING: line over 80 characters torvalds#54: FILE: mm/rmap.c:988: + * pte lock(a spinlock) is held, which implies preemtion disabled. total: 0 errors, 1 warnings, 45 lines checked ./patches/mm-add-comment-for-__mod_zone_page_stat.patch has style problems, please review. If any of these errors are false positives, please report them to the maintainer, see CHECKPATCH in MAINTAINERS. Please run checkpatch prior to sending patches Cc: Hugh Dickins <[email protected]> Cc: Jianyu Zhan <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

If an edge wake up interrupt happens during suspend path, the flow handler will mask it and set as pending. This lead to suspend sequence abort when the check for all pending edge interrupt happens. Do not disable the interrupt (wake up interrupt to CPU0 in krait SS, torvalds#54) during suspend. CRs-fixed: 583318 Change-Id: I3a5afec98d445da2dfc1f16fc20d9f3e96acc17f Signed-off-by: Murali Nalajala <[email protected]>

also fix spellos WARNING: line over 80 characters torvalds#54: FILE: mm/rmap.c:988: + * pte lock(a spinlock) is held, which implies preemtion disabled. total: 0 errors, 1 warnings, 45 lines checked ./patches/mm-add-comment-for-__mod_zone_page_stat.patch has style problems, please review. If any of these errors are false positives, please report them to the maintainer, see CHECKPATCH in MAINTAINERS. Please run checkpatch prior to sending patches Cc: Hugh Dickins <[email protected]> Cc: Jianyu Zhan <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

@progbits

It appears that gcc is better at optimising a double call to min and max rather than open coded min3 and max3. This can be observed here: $ cat min-max.c #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define min3(x, y, z) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ typeof(z) _min3 = (z); \ (void) (&_min1 == &_min2); \ (void) (&_min1 == &_min3); \ _min1 < _min2 ? (_min1 < _min3 ? _min1 : _min3) : \ (_min2 < _min3 ? _min2 : _min3); }) int fmin3(int x, int y, int z) { return min3(x, y, z); } int fmin2(int x, int y, int z) { return min(min(x, y), z); } $ gcc -O2 -o min-max.s -S min-max.c; cat min-max.s .file "min-max.c" .text .p2align 4,,15 .globl fmin3 .type fmin3, @function fmin3: .LFB0: .cfi_startproc cmpl %esi, %edi jl .L5 cmpl %esi, %edx movl %esi, %eax cmovle %edx, %eax ret .p2align 4,,10 .p2align 3 .L5: cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE0: .size fmin3, .-fmin3 .p2align 4,,15 .globl fmin2 .type fmin2, @function fmin2: .LFB1: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovle %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE1: .size fmin2, .-fmin2 .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits fmin3 function, which uses open-coded min3 macro, is compiled into total of ten instructions including a conditional branch, whereas fmin2 function, which uses two calls to min2 macro, is compiled into six instructions with no branches. Similarly, open-coded clamp produces the same code as clamp using min and max macros, but the latter is much shorter: $ cat clamp.c #define clamp(val, min, max) ({ \ typeof(val) __val = (val); \ typeof(min) __min = (min); \ typeof(max) __max = (max); \ (void) (&__val == &__min); \ (void) (&__val == &__max); \ __val = __val < __min ? __min: __val; \ __val > __max ? __max: __val; }) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define max(x, y) ({ \ typeof(x) _max1 = (x); \ typeof(y) _max2 = (y); \ (void) (&_max1 == &_max2); \ _max1 > _max2 ? _max1 : _max2; }) int fclamp(int v, int min, int max) { return clamp(v, min, max); } int fclampmm(int v, int min, int max) { return min(max(v, min), max); } $ gcc -O2 -o clamp.s -S clamp.c; cat clamp.s .file "clamp.c" .text .p2align 4,,15 .globl fclamp .type fclamp, @function fclamp: .LFB0: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovge %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE0: .size fclamp, .-fclamp .p2align 4,,15 .globl fclampmm .type fclampmm, @function fclampmm: .LFB1: .cfi_startproc cmpl %edi, %esi cmovge %esi, %edi cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE1: .size fclampmm, .-fclampmm .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits Linux mpn-glaptop 3.13.0-29-generic #53~precise1-Ubuntu SMP Wed Jun 4 22:06:25 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 51224656 Jun 17 14:15 vmlinux.before -rwx------ 1 mpn eng 51224608 Jun 17 13:57 vmlinux.after 48 bytes reduction. The do_fault_around was a few instruction shorter and as far as I can tell saved 12 bytes on the stack, i.e.: $ grep -e rsp -e pop -e push do_fault_around.* do_fault_around.before.s:push %rbp do_fault_around.before.s:mov %rsp,%rbp do_fault_around.before.s:push %r13 do_fault_around.before.s:push %r12 do_fault_around.before.s:push %rbx do_fault_around.before.s:sub $0x38,%rsp do_fault_around.before.s:add $0x38,%rsp do_fault_around.before.s:pop %rbx do_fault_around.before.s:pop %r12 do_fault_around.before.s:pop %r13 do_fault_around.before.s:pop %rbp do_fault_around.after.s:push %rbp do_fault_around.after.s:mov %rsp,%rbp do_fault_around.after.s:push %r12 do_fault_around.after.s:push %rbx do_fault_around.after.s:sub $0x30,%rsp do_fault_around.after.s:add $0x30,%rsp do_fault_around.after.s:pop %rbx do_fault_around.after.s:pop %r12 do_fault_around.after.s:pop %rbp or here side-by-side: Before After push %rbp push %rbp mov %rsp,%rbp mov %rsp,%rbp push %r13 push %r12 push %r12 push %rbx push %rbx sub $0x38,%rsp sub $0x30,%rsp add $0x38,%rsp add $0x30,%rsp pop %rbx pop %rbx pop %r12 pop %r12 pop %r13 pop %rbp pop %rbp There are also fewer branches: $ grep ^j do_fault_around.* do_fault_around.before.s:jae ffffffff812079b7 do_fault_around.before.s:jmp ffffffff812079c5 do_fault_around.before.s:jmp ffffffff81207a14 do_fault_around.before.s:ja ffffffff812079f9 do_fault_around.before.s:jb ffffffff81207a10 do_fault_around.before.s:jmp ffffffff81207a63 do_fault_around.before.s:jne ffffffff812079df do_fault_around.after.s:jmp ffffffff812079fd do_fault_around.after.s:ja ffffffff812079e2 do_fault_around.after.s:jb ffffffff812079f9 do_fault_around.after.s:jmp ffffffff81207a4c do_fault_around.after.s:jne ffffffff812079c8 And here's with allyesconfig on a different machine: $ uname -a; gcc --version; ls -l vmlinux.* Linux erwin 3.14.7-mn torvalds#54 SMP Sun Jun 15 11:25:08 CEST 2014 x86_64 AMD Phenom(tm) II X3 710 Processor AuthenticAMD GNU/Linux gcc (GCC) 4.8.3 Copyright (C) 2013 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mina86 mina86 230616126 Jun 17 15:39 vmlinux.before* -rwx------ 1 mina86 mina86 230614861 Jun 17 14:36 vmlinux.after* 1265 bytes reduction. Signed-off-by: Michal Nazarewicz <[email protected]> Signed-off-by: Hagen Paul Pfeifer <[email protected]> Acked-by: Steven Rostedt <[email protected]> Cc: Hagen Paul Pfeifer <[email protected]> Cc: David Rientjes <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

@progbits

It appears that gcc is better at optimising a double call to min and max rather than open coded min3 and max3. This can be observed here: $ cat min-max.c #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define min3(x, y, z) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ typeof(z) _min3 = (z); \ (void) (&_min1 == &_min2); \ (void) (&_min1 == &_min3); \ _min1 < _min2 ? (_min1 < _min3 ? _min1 : _min3) : \ (_min2 < _min3 ? _min2 : _min3); }) int fmin3(int x, int y, int z) { return min3(x, y, z); } int fmin2(int x, int y, int z) { return min(min(x, y), z); } $ gcc -O2 -o min-max.s -S min-max.c; cat min-max.s .file "min-max.c" .text .p2align 4,,15 .globl fmin3 .type fmin3, @function fmin3: .LFB0: .cfi_startproc cmpl %esi, %edi jl .L5 cmpl %esi, %edx movl %esi, %eax cmovle %edx, %eax ret .p2align 4,,10 .p2align 3 .L5: cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE0: .size fmin3, .-fmin3 .p2align 4,,15 .globl fmin2 .type fmin2, @function fmin2: .LFB1: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovle %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE1: .size fmin2, .-fmin2 .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits fmin3 function, which uses open-coded min3 macro, is compiled into total of ten instructions including a conditional branch, whereas fmin2 function, which uses two calls to min2 macro, is compiled into six instructions with no branches. Similarly, open-coded clamp produces the same code as clamp using min and max macros, but the latter is much shorter: $ cat clamp.c #define clamp(val, min, max) ({ \ typeof(val) __val = (val); \ typeof(min) __min = (min); \ typeof(max) __max = (max); \ (void) (&__val == &__min); \ (void) (&__val == &__max); \ __val = __val < __min ? __min: __val; \ __val > __max ? __max: __val; }) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define max(x, y) ({ \ typeof(x) _max1 = (x); \ typeof(y) _max2 = (y); \ (void) (&_max1 == &_max2); \ _max1 > _max2 ? _max1 : _max2; }) int fclamp(int v, int min, int max) { return clamp(v, min, max); } int fclampmm(int v, int min, int max) { return min(max(v, min), max); } $ gcc -O2 -o clamp.s -S clamp.c; cat clamp.s .file "clamp.c" .text .p2align 4,,15 .globl fclamp .type fclamp, @function fclamp: .LFB0: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovge %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE0: .size fclamp, .-fclamp .p2align 4,,15 .globl fclampmm .type fclampmm, @function fclampmm: .LFB1: .cfi_startproc cmpl %edi, %esi cmovge %esi, %edi cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE1: .size fclampmm, .-fclampmm .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits Linux mpn-glaptop 3.13.0-29-generic #53~precise1-Ubuntu SMP Wed Jun 4 22:06:25 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 51224656 Jun 17 14:15 vmlinux.before -rwx------ 1 mpn eng 51224608 Jun 17 13:57 vmlinux.after 48 bytes reduction. The do_fault_around was a few instruction shorter and as far as I can tell saved 12 bytes on the stack, i.e.: $ grep -e rsp -e pop -e push do_fault_around.* do_fault_around.before.s:push %rbp do_fault_around.before.s:mov %rsp,%rbp do_fault_around.before.s:push %r13 do_fault_around.before.s:push %r12 do_fault_around.before.s:push %rbx do_fault_around.before.s:sub $0x38,%rsp do_fault_around.before.s:add $0x38,%rsp do_fault_around.before.s:pop %rbx do_fault_around.before.s:pop %r12 do_fault_around.before.s:pop %r13 do_fault_around.before.s:pop %rbp do_fault_around.after.s:push %rbp do_fault_around.after.s:mov %rsp,%rbp do_fault_around.after.s:push %r12 do_fault_around.after.s:push %rbx do_fault_around.after.s:sub $0x30,%rsp do_fault_around.after.s:add $0x30,%rsp do_fault_around.after.s:pop %rbx do_fault_around.after.s:pop %r12 do_fault_around.after.s:pop %rbp or here side-by-side: Before After push %rbp push %rbp mov %rsp,%rbp mov %rsp,%rbp push %r13 push %r12 push %r12 push %rbx push %rbx sub $0x38,%rsp sub $0x30,%rsp add $0x38,%rsp add $0x30,%rsp pop %rbx pop %rbx pop %r12 pop %r12 pop %r13 pop %rbp pop %rbp There are also fewer branches: $ grep ^j do_fault_around.* do_fault_around.before.s:jae ffffffff812079b7 do_fault_around.before.s:jmp ffffffff812079c5 do_fault_around.before.s:jmp ffffffff81207a14 do_fault_around.before.s:ja ffffffff812079f9 do_fault_around.before.s:jb ffffffff81207a10 do_fault_around.before.s:jmp ffffffff81207a63 do_fault_around.before.s:jne ffffffff812079df do_fault_around.after.s:jmp ffffffff812079fd do_fault_around.after.s:ja ffffffff812079e2 do_fault_around.after.s:jb ffffffff812079f9 do_fault_around.after.s:jmp ffffffff81207a4c do_fault_around.after.s:jne ffffffff812079c8 And here's with allyesconfig on a different machine: $ uname -a; gcc --version; ls -l vmlinux.* Linux erwin 3.14.7-mn torvalds#54 SMP Sun Jun 15 11:25:08 CEST 2014 x86_64 AMD Phenom(tm) II X3 710 Processor AuthenticAMD GNU/Linux gcc (GCC) 4.8.3 Copyright (C) 2013 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 437027411 Jun 20 16:04 vmlinux.before -rwx------ 1 mpn eng 437026881 Jun 20 15:30 vmlinux.after 530 bytes reduction. Signed-off-by: Michal Nazarewicz <[email protected]> Signed-off-by: Hagen Paul Pfeifer <[email protected]> Acked-by: Steven Rostedt <[email protected]> Cc: Hagen Paul Pfeifer <[email protected]> Cc: David Rientjes <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

@progbits

It appears that gcc is better at optimising a double call to min and max rather than open coded min3 and max3. This can be observed here: $ cat min-max.c #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define min3(x, y, z) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ typeof(z) _min3 = (z); \ (void) (&_min1 == &_min2); \ (void) (&_min1 == &_min3); \ _min1 < _min2 ? (_min1 < _min3 ? _min1 : _min3) : \ (_min2 < _min3 ? _min2 : _min3); }) int fmin3(int x, int y, int z) { return min3(x, y, z); } int fmin2(int x, int y, int z) { return min(min(x, y), z); } $ gcc -O2 -o min-max.s -S min-max.c; cat min-max.s .file "min-max.c" .text .p2align 4,,15 .globl fmin3 .type fmin3, @function fmin3: .LFB0: .cfi_startproc cmpl %esi, %edi jl .L5 cmpl %esi, %edx movl %esi, %eax cmovle %edx, %eax ret .p2align 4,,10 .p2align 3 .L5: cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE0: .size fmin3, .-fmin3 .p2align 4,,15 .globl fmin2 .type fmin2, @function fmin2: .LFB1: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovle %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE1: .size fmin2, .-fmin2 .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits fmin3 function, which uses open-coded min3 macro, is compiled into total of ten instructions including a conditional branch, whereas fmin2 function, which uses two calls to min2 macro, is compiled into six instructions with no branches. Similarly, open-coded clamp produces the same code as clamp using min and max macros, but the latter is much shorter: $ cat clamp.c #define clamp(val, min, max) ({ \ typeof(val) __val = (val); \ typeof(min) __min = (min); \ typeof(max) __max = (max); \ (void) (&__val == &__min); \ (void) (&__val == &__max); \ __val = __val < __min ? __min: __val; \ __val > __max ? __max: __val; }) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define max(x, y) ({ \ typeof(x) _max1 = (x); \ typeof(y) _max2 = (y); \ (void) (&_max1 == &_max2); \ _max1 > _max2 ? _max1 : _max2; }) int fclamp(int v, int min, int max) { return clamp(v, min, max); } int fclampmm(int v, int min, int max) { return min(max(v, min), max); } $ gcc -O2 -o clamp.s -S clamp.c; cat clamp.s .file "clamp.c" .text .p2align 4,,15 .globl fclamp .type fclamp, @function fclamp: .LFB0: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovge %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE0: .size fclamp, .-fclamp .p2align 4,,15 .globl fclampmm .type fclampmm, @function fclampmm: .LFB1: .cfi_startproc cmpl %edi, %esi cmovge %esi, %edi cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE1: .size fclampmm, .-fclampmm .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits Linux mpn-glaptop 3.13.0-29-generic #53~precise1-Ubuntu SMP Wed Jun 4 22:06:25 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 51224656 Jun 17 14:15 vmlinux.before -rwx------ 1 mpn eng 51224608 Jun 17 13:57 vmlinux.after 48 bytes reduction. The do_fault_around was a few instruction shorter and as far as I can tell saved 12 bytes on the stack, i.e.: $ grep -e rsp -e pop -e push do_fault_around.* do_fault_around.before.s:push %rbp do_fault_around.before.s:mov %rsp,%rbp do_fault_around.before.s:push %r13 do_fault_around.before.s:push %r12 do_fault_around.before.s:push %rbx do_fault_around.before.s:sub $0x38,%rsp do_fault_around.before.s:add $0x38,%rsp do_fault_around.before.s:pop %rbx do_fault_around.before.s:pop %r12 do_fault_around.before.s:pop %r13 do_fault_around.before.s:pop %rbp do_fault_around.after.s:push %rbp do_fault_around.after.s:mov %rsp,%rbp do_fault_around.after.s:push %r12 do_fault_around.after.s:push %rbx do_fault_around.after.s:sub $0x30,%rsp do_fault_around.after.s:add $0x30,%rsp do_fault_around.after.s:pop %rbx do_fault_around.after.s:pop %r12 do_fault_around.after.s:pop %rbp or here side-by-side: Before After push %rbp push %rbp mov %rsp,%rbp mov %rsp,%rbp push %r13 push %r12 push %r12 push %rbx push %rbx sub $0x38,%rsp sub $0x30,%rsp add $0x38,%rsp add $0x30,%rsp pop %rbx pop %rbx pop %r12 pop %r12 pop %r13 pop %rbp pop %rbp There are also fewer branches: $ grep ^j do_fault_around.* do_fault_around.before.s:jae ffffffff812079b7 do_fault_around.before.s:jmp ffffffff812079c5 do_fault_around.before.s:jmp ffffffff81207a14 do_fault_around.before.s:ja ffffffff812079f9 do_fault_around.before.s:jb ffffffff81207a10 do_fault_around.before.s:jmp ffffffff81207a63 do_fault_around.before.s:jne ffffffff812079df do_fault_around.after.s:jmp ffffffff812079fd do_fault_around.after.s:ja ffffffff812079e2 do_fault_around.after.s:jb ffffffff812079f9 do_fault_around.after.s:jmp ffffffff81207a4c do_fault_around.after.s:jne ffffffff812079c8 And here's with allyesconfig on a different machine: $ uname -a; gcc --version; ls -l vmlinux.* Linux erwin 3.14.7-mn torvalds#54 SMP Sun Jun 15 11:25:08 CEST 2014 x86_64 AMD Phenom(tm) II X3 710 Processor AuthenticAMD GNU/Linux gcc (GCC) 4.8.3 Copyright (C) 2013 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 437027411 Jun 20 16:04 vmlinux.before -rwx------ 1 mpn eng 437026881 Jun 20 15:30 vmlinux.after 530 bytes reduction. Signed-off-by: Michal Nazarewicz <[email protected]> Signed-off-by: Hagen Paul Pfeifer <[email protected]> Acked-by: Steven Rostedt <[email protected]> Cc: Hagen Paul Pfeifer <[email protected]> Cc: David Rientjes <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

@progbits

It appears that gcc is better at optimising a double call to min and max rather than open coded min3 and max3. This can be observed here: $ cat min-max.c #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define min3(x, y, z) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ typeof(z) _min3 = (z); \ (void) (&_min1 == &_min2); \ (void) (&_min1 == &_min3); \ _min1 < _min2 ? (_min1 < _min3 ? _min1 : _min3) : \ (_min2 < _min3 ? _min2 : _min3); }) int fmin3(int x, int y, int z) { return min3(x, y, z); } int fmin2(int x, int y, int z) { return min(min(x, y), z); } $ gcc -O2 -o min-max.s -S min-max.c; cat min-max.s .file "min-max.c" .text .p2align 4,,15 .globl fmin3 .type fmin3, @function fmin3: .LFB0: .cfi_startproc cmpl %esi, %edi jl .L5 cmpl %esi, %edx movl %esi, %eax cmovle %edx, %eax ret .p2align 4,,10 .p2align 3 .L5: cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE0: .size fmin3, .-fmin3 .p2align 4,,15 .globl fmin2 .type fmin2, @function fmin2: .LFB1: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovle %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE1: .size fmin2, .-fmin2 .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits fmin3 function, which uses open-coded min3 macro, is compiled into total of ten instructions including a conditional branch, whereas fmin2 function, which uses two calls to min2 macro, is compiled into six instructions with no branches. Similarly, open-coded clamp produces the same code as clamp using min and max macros, but the latter is much shorter: $ cat clamp.c #define clamp(val, min, max) ({ \ typeof(val) __val = (val); \ typeof(min) __min = (min); \ typeof(max) __max = (max); \ (void) (&__val == &__min); \ (void) (&__val == &__max); \ __val = __val < __min ? __min: __val; \ __val > __max ? __max: __val; }) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define max(x, y) ({ \ typeof(x) _max1 = (x); \ typeof(y) _max2 = (y); \ (void) (&_max1 == &_max2); \ _max1 > _max2 ? _max1 : _max2; }) int fclamp(int v, int min, int max) { return clamp(v, min, max); } int fclampmm(int v, int min, int max) { return min(max(v, min), max); } $ gcc -O2 -o clamp.s -S clamp.c; cat clamp.s .file "clamp.c" .text .p2align 4,,15 .globl fclamp .type fclamp, @function fclamp: .LFB0: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovge %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE0: .size fclamp, .-fclamp .p2align 4,,15 .globl fclampmm .type fclampmm, @function fclampmm: .LFB1: .cfi_startproc cmpl %edi, %esi cmovge %esi, %edi cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE1: .size fclampmm, .-fclampmm .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits Linux mpn-glaptop 3.13.0-29-generic #53~precise1-Ubuntu SMP Wed Jun 4 22:06:25 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 51224656 Jun 17 14:15 vmlinux.before -rwx------ 1 mpn eng 51224608 Jun 17 13:57 vmlinux.after 48 bytes reduction. The do_fault_around was a few instruction shorter and as far as I can tell saved 12 bytes on the stack, i.e.: $ grep -e rsp -e pop -e push do_fault_around.* do_fault_around.before.s:push %rbp do_fault_around.before.s:mov %rsp,%rbp do_fault_around.before.s:push %r13 do_fault_around.before.s:push %r12 do_fault_around.before.s:push %rbx do_fault_around.before.s:sub $0x38,%rsp do_fault_around.before.s:add $0x38,%rsp do_fault_around.before.s:pop %rbx do_fault_around.before.s:pop %r12 do_fault_around.before.s:pop %r13 do_fault_around.before.s:pop %rbp do_fault_around.after.s:push %rbp do_fault_around.after.s:mov %rsp,%rbp do_fault_around.after.s:push %r12 do_fault_around.after.s:push %rbx do_fault_around.after.s:sub $0x30,%rsp do_fault_around.after.s:add $0x30,%rsp do_fault_around.after.s:pop %rbx do_fault_around.after.s:pop %r12 do_fault_around.after.s:pop %rbp or here side-by-side: Before After push %rbp push %rbp mov %rsp,%rbp mov %rsp,%rbp push %r13 push %r12 push %r12 push %rbx push %rbx sub $0x38,%rsp sub $0x30,%rsp add $0x38,%rsp add $0x30,%rsp pop %rbx pop %rbx pop %r12 pop %r12 pop %r13 pop %rbp pop %rbp There are also fewer branches: $ grep ^j do_fault_around.* do_fault_around.before.s:jae ffffffff812079b7 do_fault_around.before.s:jmp ffffffff812079c5 do_fault_around.before.s:jmp ffffffff81207a14 do_fault_around.before.s:ja ffffffff812079f9 do_fault_around.before.s:jb ffffffff81207a10 do_fault_around.before.s:jmp ffffffff81207a63 do_fault_around.before.s:jne ffffffff812079df do_fault_around.after.s:jmp ffffffff812079fd do_fault_around.after.s:ja ffffffff812079e2 do_fault_around.after.s:jb ffffffff812079f9 do_fault_around.after.s:jmp ffffffff81207a4c do_fault_around.after.s:jne ffffffff812079c8 And here's with allyesconfig on a different machine: $ uname -a; gcc --version; ls -l vmlinux.* Linux erwin 3.14.7-mn torvalds#54 SMP Sun Jun 15 11:25:08 CEST 2014 x86_64 AMD Phenom(tm) II X3 710 Processor AuthenticAMD GNU/Linux gcc (GCC) 4.8.3 Copyright (C) 2013 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 437027411 Jun 20 16:04 vmlinux.before -rwx------ 1 mpn eng 437026881 Jun 20 15:30 vmlinux.after 530 bytes reduction. Signed-off-by: Michal Nazarewicz <[email protected]> Signed-off-by: Hagen Paul Pfeifer <[email protected]> Acked-by: Steven Rostedt <[email protected]> Cc: Hagen Paul Pfeifer <[email protected]> Cc: David Rientjes <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

@progbits

It appears that gcc is better at optimising a double call to min and max rather than open coded min3 and max3. This can be observed here: $ cat min-max.c #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define min3(x, y, z) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ typeof(z) _min3 = (z); \ (void) (&_min1 == &_min2); \ (void) (&_min1 == &_min3); \ _min1 < _min2 ? (_min1 < _min3 ? _min1 : _min3) : \ (_min2 < _min3 ? _min2 : _min3); }) int fmin3(int x, int y, int z) { return min3(x, y, z); } int fmin2(int x, int y, int z) { return min(min(x, y), z); } $ gcc -O2 -o min-max.s -S min-max.c; cat min-max.s .file "min-max.c" .text .p2align 4,,15 .globl fmin3 .type fmin3, @function fmin3: .LFB0: .cfi_startproc cmpl %esi, %edi jl .L5 cmpl %esi, %edx movl %esi, %eax cmovle %edx, %eax ret .p2align 4,,10 .p2align 3 .L5: cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE0: .size fmin3, .-fmin3 .p2align 4,,15 .globl fmin2 .type fmin2, @function fmin2: .LFB1: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovle %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE1: .size fmin2, .-fmin2 .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits fmin3 function, which uses open-coded min3 macro, is compiled into total of ten instructions including a conditional branch, whereas fmin2 function, which uses two calls to min2 macro, is compiled into six instructions with no branches. Similarly, open-coded clamp produces the same code as clamp using min and max macros, but the latter is much shorter: $ cat clamp.c #define clamp(val, min, max) ({ \ typeof(val) __val = (val); \ typeof(min) __min = (min); \ typeof(max) __max = (max); \ (void) (&__val == &__min); \ (void) (&__val == &__max); \ __val = __val < __min ? __min: __val; \ __val > __max ? __max: __val; }) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define max(x, y) ({ \ typeof(x) _max1 = (x); \ typeof(y) _max2 = (y); \ (void) (&_max1 == &_max2); \ _max1 > _max2 ? _max1 : _max2; }) int fclamp(int v, int min, int max) { return clamp(v, min, max); } int fclampmm(int v, int min, int max) { return min(max(v, min), max); } $ gcc -O2 -o clamp.s -S clamp.c; cat clamp.s .file "clamp.c" .text .p2align 4,,15 .globl fclamp .type fclamp, @function fclamp: .LFB0: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovge %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE0: .size fclamp, .-fclamp .p2align 4,,15 .globl fclampmm .type fclampmm, @function fclampmm: .LFB1: .cfi_startproc cmpl %edi, %esi cmovge %esi, %edi cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE1: .size fclampmm, .-fclampmm .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits Linux mpn-glaptop 3.13.0-29-generic #53~precise1-Ubuntu SMP Wed Jun 4 22:06:25 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 51224656 Jun 17 14:15 vmlinux.before -rwx------ 1 mpn eng 51224608 Jun 17 13:57 vmlinux.after 48 bytes reduction. The do_fault_around was a few instruction shorter and as far as I can tell saved 12 bytes on the stack, i.e.: $ grep -e rsp -e pop -e push do_fault_around.* do_fault_around.before.s:push %rbp do_fault_around.before.s:mov %rsp,%rbp do_fault_around.before.s:push %r13 do_fault_around.before.s:push %r12 do_fault_around.before.s:push %rbx do_fault_around.before.s:sub $0x38,%rsp do_fault_around.before.s:add $0x38,%rsp do_fault_around.before.s:pop %rbx do_fault_around.before.s:pop %r12 do_fault_around.before.s:pop %r13 do_fault_around.before.s:pop %rbp do_fault_around.after.s:push %rbp do_fault_around.after.s:mov %rsp,%rbp do_fault_around.after.s:push %r12 do_fault_around.after.s:push %rbx do_fault_around.after.s:sub $0x30,%rsp do_fault_around.after.s:add $0x30,%rsp do_fault_around.after.s:pop %rbx do_fault_around.after.s:pop %r12 do_fault_around.after.s:pop %rbp or here side-by-side: Before After push %rbp push %rbp mov %rsp,%rbp mov %rsp,%rbp push %r13 push %r12 push %r12 push %rbx push %rbx sub $0x38,%rsp sub $0x30,%rsp add $0x38,%rsp add $0x30,%rsp pop %rbx pop %rbx pop %r12 pop %r12 pop %r13 pop %rbp pop %rbp There are also fewer branches: $ grep ^j do_fault_around.* do_fault_around.before.s:jae ffffffff812079b7 do_fault_around.before.s:jmp ffffffff812079c5 do_fault_around.before.s:jmp ffffffff81207a14 do_fault_around.before.s:ja ffffffff812079f9 do_fault_around.before.s:jb ffffffff81207a10 do_fault_around.before.s:jmp ffffffff81207a63 do_fault_around.before.s:jne ffffffff812079df do_fault_around.after.s:jmp ffffffff812079fd do_fault_around.after.s:ja ffffffff812079e2 do_fault_around.after.s:jb ffffffff812079f9 do_fault_around.after.s:jmp ffffffff81207a4c do_fault_around.after.s:jne ffffffff812079c8 And here's with allyesconfig on a different machine: $ uname -a; gcc --version; ls -l vmlinux.* Linux erwin 3.14.7-mn torvalds#54 SMP Sun Jun 15 11:25:08 CEST 2014 x86_64 AMD Phenom(tm) II X3 710 Processor AuthenticAMD GNU/Linux gcc (GCC) 4.8.3 Copyright (C) 2013 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 437027411 Jun 20 16:04 vmlinux.before -rwx------ 1 mpn eng 437026881 Jun 20 15:30 vmlinux.after 530 bytes reduction. Signed-off-by: Michal Nazarewicz <[email protected]> Signed-off-by: Hagen Paul Pfeifer <[email protected]> Acked-by: Steven Rostedt <[email protected]> Cc: Hagen Paul Pfeifer <[email protected]> Cc: David Rientjes <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

@progbits

It appears that gcc is better at optimising a double call to min and max rather than open coded min3 and max3. This can be observed here: $ cat min-max.c #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define min3(x, y, z) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ typeof(z) _min3 = (z); \ (void) (&_min1 == &_min2); \ (void) (&_min1 == &_min3); \ _min1 < _min2 ? (_min1 < _min3 ? _min1 : _min3) : \ (_min2 < _min3 ? _min2 : _min3); }) int fmin3(int x, int y, int z) { return min3(x, y, z); } int fmin2(int x, int y, int z) { return min(min(x, y), z); } $ gcc -O2 -o min-max.s -S min-max.c; cat min-max.s .file "min-max.c" .text .p2align 4,,15 .globl fmin3 .type fmin3, @function fmin3: .LFB0: .cfi_startproc cmpl %esi, %edi jl .L5 cmpl %esi, %edx movl %esi, %eax cmovle %edx, %eax ret .p2align 4,,10 .p2align 3 .L5: cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE0: .size fmin3, .-fmin3 .p2align 4,,15 .globl fmin2 .type fmin2, @function fmin2: .LFB1: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovle %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE1: .size fmin2, .-fmin2 .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits fmin3 function, which uses open-coded min3 macro, is compiled into total of ten instructions including a conditional branch, whereas fmin2 function, which uses two calls to min2 macro, is compiled into six instructions with no branches. Similarly, open-coded clamp produces the same code as clamp using min and max macros, but the latter is much shorter: $ cat clamp.c #define clamp(val, min, max) ({ \ typeof(val) __val = (val); \ typeof(min) __min = (min); \ typeof(max) __max = (max); \ (void) (&__val == &__min); \ (void) (&__val == &__max); \ __val = __val < __min ? __min: __val; \ __val > __max ? __max: __val; }) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define max(x, y) ({ \ typeof(x) _max1 = (x); \ typeof(y) _max2 = (y); \ (void) (&_max1 == &_max2); \ _max1 > _max2 ? _max1 : _max2; }) int fclamp(int v, int min, int max) { return clamp(v, min, max); } int fclampmm(int v, int min, int max) { return min(max(v, min), max); } $ gcc -O2 -o clamp.s -S clamp.c; cat clamp.s .file "clamp.c" .text .p2align 4,,15 .globl fclamp .type fclamp, @function fclamp: .LFB0: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovge %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE0: .size fclamp, .-fclamp .p2align 4,,15 .globl fclampmm .type fclampmm, @function fclampmm: .LFB1: .cfi_startproc cmpl %edi, %esi cmovge %esi, %edi cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE1: .size fclampmm, .-fclampmm .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits Linux mpn-glaptop 3.13.0-29-generic #53~precise1-Ubuntu SMP Wed Jun 4 22:06:25 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 51224656 Jun 17 14:15 vmlinux.before -rwx------ 1 mpn eng 51224608 Jun 17 13:57 vmlinux.after 48 bytes reduction. The do_fault_around was a few instruction shorter and as far as I can tell saved 12 bytes on the stack, i.e.: $ grep -e rsp -e pop -e push do_fault_around.* do_fault_around.before.s:push %rbp do_fault_around.before.s:mov %rsp,%rbp do_fault_around.before.s:push %r13 do_fault_around.before.s:push %r12 do_fault_around.before.s:push %rbx do_fault_around.before.s:sub $0x38,%rsp do_fault_around.before.s:add $0x38,%rsp do_fault_around.before.s:pop %rbx do_fault_around.before.s:pop %r12 do_fault_around.before.s:pop %r13 do_fault_around.before.s:pop %rbp do_fault_around.after.s:push %rbp do_fault_around.after.s:mov %rsp,%rbp do_fault_around.after.s:push %r12 do_fault_around.after.s:push %rbx do_fault_around.after.s:sub $0x30,%rsp do_fault_around.after.s:add $0x30,%rsp do_fault_around.after.s:pop %rbx do_fault_around.after.s:pop %r12 do_fault_around.after.s:pop %rbp or here side-by-side: Before After push %rbp push %rbp mov %rsp,%rbp mov %rsp,%rbp push %r13 push %r12 push %r12 push %rbx push %rbx sub $0x38,%rsp sub $0x30,%rsp add $0x38,%rsp add $0x30,%rsp pop %rbx pop %rbx pop %r12 pop %r12 pop %r13 pop %rbp pop %rbp There are also fewer branches: $ grep ^j do_fault_around.* do_fault_around.before.s:jae ffffffff812079b7 do_fault_around.before.s:jmp ffffffff812079c5 do_fault_around.before.s:jmp ffffffff81207a14 do_fault_around.before.s:ja ffffffff812079f9 do_fault_around.before.s:jb ffffffff81207a10 do_fault_around.before.s:jmp ffffffff81207a63 do_fault_around.before.s:jne ffffffff812079df do_fault_around.after.s:jmp ffffffff812079fd do_fault_around.after.s:ja ffffffff812079e2 do_fault_around.after.s:jb ffffffff812079f9 do_fault_around.after.s:jmp ffffffff81207a4c do_fault_around.after.s:jne ffffffff812079c8 And here's with allyesconfig on a different machine: $ uname -a; gcc --version; ls -l vmlinux.* Linux erwin 3.14.7-mn torvalds#54 SMP Sun Jun 15 11:25:08 CEST 2014 x86_64 AMD Phenom(tm) II X3 710 Processor AuthenticAMD GNU/Linux gcc (GCC) 4.8.3 Copyright (C) 2013 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 437027411 Jun 20 16:04 vmlinux.before -rwx------ 1 mpn eng 437026881 Jun 20 15:30 vmlinux.after 530 bytes reduction. Signed-off-by: Michal Nazarewicz <[email protected]> Signed-off-by: Hagen Paul Pfeifer <[email protected]> Acked-by: Steven Rostedt <[email protected]> Cc: Hagen Paul Pfeifer <[email protected]> Cc: David Rientjes <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

@progbits

It appears that gcc is better at optimising a double call to min and max rather than open coded min3 and max3. This can be observed here: $ cat min-max.c #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define min3(x, y, z) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ typeof(z) _min3 = (z); \ (void) (&_min1 == &_min2); \ (void) (&_min1 == &_min3); \ _min1 < _min2 ? (_min1 < _min3 ? _min1 : _min3) : \ (_min2 < _min3 ? _min2 : _min3); }) int fmin3(int x, int y, int z) { return min3(x, y, z); } int fmin2(int x, int y, int z) { return min(min(x, y), z); } $ gcc -O2 -o min-max.s -S min-max.c; cat min-max.s .file "min-max.c" .text .p2align 4,,15 .globl fmin3 .type fmin3, @function fmin3: .LFB0: .cfi_startproc cmpl %esi, %edi jl .L5 cmpl %esi, %edx movl %esi, %eax cmovle %edx, %eax ret .p2align 4,,10 .p2align 3 .L5: cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE0: .size fmin3, .-fmin3 .p2align 4,,15 .globl fmin2 .type fmin2, @function fmin2: .LFB1: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovle %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE1: .size fmin2, .-fmin2 .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits fmin3 function, which uses open-coded min3 macro, is compiled into total of ten instructions including a conditional branch, whereas fmin2 function, which uses two calls to min2 macro, is compiled into six instructions with no branches. Similarly, open-coded clamp produces the same code as clamp using min and max macros, but the latter is much shorter: $ cat clamp.c #define clamp(val, min, max) ({ \ typeof(val) __val = (val); \ typeof(min) __min = (min); \ typeof(max) __max = (max); \ (void) (&__val == &__min); \ (void) (&__val == &__max); \ __val = __val < __min ? __min: __val; \ __val > __max ? __max: __val; }) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define max(x, y) ({ \ typeof(x) _max1 = (x); \ typeof(y) _max2 = (y); \ (void) (&_max1 == &_max2); \ _max1 > _max2 ? _max1 : _max2; }) int fclamp(int v, int min, int max) { return clamp(v, min, max); } int fclampmm(int v, int min, int max) { return min(max(v, min), max); } $ gcc -O2 -o clamp.s -S clamp.c; cat clamp.s .file "clamp.c" .text .p2align 4,,15 .globl fclamp .type fclamp, @function fclamp: .LFB0: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovge %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE0: .size fclamp, .-fclamp .p2align 4,,15 .globl fclampmm .type fclampmm, @function fclampmm: .LFB1: .cfi_startproc cmpl %edi, %esi cmovge %esi, %edi cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE1: .size fclampmm, .-fclampmm .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits Linux mpn-glaptop 3.13.0-29-generic #53~precise1-Ubuntu SMP Wed Jun 4 22:06:25 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 51224656 Jun 17 14:15 vmlinux.before -rwx------ 1 mpn eng 51224608 Jun 17 13:57 vmlinux.after 48 bytes reduction. The do_fault_around was a few instruction shorter and as far as I can tell saved 12 bytes on the stack, i.e.: $ grep -e rsp -e pop -e push do_fault_around.* do_fault_around.before.s:push %rbp do_fault_around.before.s:mov %rsp,%rbp do_fault_around.before.s:push %r13 do_fault_around.before.s:push %r12 do_fault_around.before.s:push %rbx do_fault_around.before.s:sub $0x38,%rsp do_fault_around.before.s:add $0x38,%rsp do_fault_around.before.s:pop %rbx do_fault_around.before.s:pop %r12 do_fault_around.before.s:pop %r13 do_fault_around.before.s:pop %rbp do_fault_around.after.s:push %rbp do_fault_around.after.s:mov %rsp,%rbp do_fault_around.after.s:push %r12 do_fault_around.after.s:push %rbx do_fault_around.after.s:sub $0x30,%rsp do_fault_around.after.s:add $0x30,%rsp do_fault_around.after.s:pop %rbx do_fault_around.after.s:pop %r12 do_fault_around.after.s:pop %rbp or here side-by-side: Before After push %rbp push %rbp mov %rsp,%rbp mov %rsp,%rbp push %r13 push %r12 push %r12 push %rbx push %rbx sub $0x38,%rsp sub $0x30,%rsp add $0x38,%rsp add $0x30,%rsp pop %rbx pop %rbx pop %r12 pop %r12 pop %r13 pop %rbp pop %rbp There are also fewer branches: $ grep ^j do_fault_around.* do_fault_around.before.s:jae ffffffff812079b7 do_fault_around.before.s:jmp ffffffff812079c5 do_fault_around.before.s:jmp ffffffff81207a14 do_fault_around.before.s:ja ffffffff812079f9 do_fault_around.before.s:jb ffffffff81207a10 do_fault_around.before.s:jmp ffffffff81207a63 do_fault_around.before.s:jne ffffffff812079df do_fault_around.after.s:jmp ffffffff812079fd do_fault_around.after.s:ja ffffffff812079e2 do_fault_around.after.s:jb ffffffff812079f9 do_fault_around.after.s:jmp ffffffff81207a4c do_fault_around.after.s:jne ffffffff812079c8 And here's with allyesconfig on a different machine: $ uname -a; gcc --version; ls -l vmlinux.* Linux erwin 3.14.7-mn torvalds#54 SMP Sun Jun 15 11:25:08 CEST 2014 x86_64 AMD Phenom(tm) II X3 710 Processor AuthenticAMD GNU/Linux gcc (GCC) 4.8.3 Copyright (C) 2013 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 437027411 Jun 20 16:04 vmlinux.before -rwx------ 1 mpn eng 437026881 Jun 20 15:30 vmlinux.after 530 bytes reduction. Signed-off-by: Michal Nazarewicz <[email protected]> Signed-off-by: Hagen Paul Pfeifer <[email protected]> Acked-by: Steven Rostedt <[email protected]> Cc: Hagen Paul Pfeifer <[email protected]> Cc: David Rientjes <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

@progbits

It appears that gcc is better at optimising a double call to min and max rather than open coded min3 and max3. This can be observed here: $ cat min-max.c #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define min3(x, y, z) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ typeof(z) _min3 = (z); \ (void) (&_min1 == &_min2); \ (void) (&_min1 == &_min3); \ _min1 < _min2 ? (_min1 < _min3 ? _min1 : _min3) : \ (_min2 < _min3 ? _min2 : _min3); }) int fmin3(int x, int y, int z) { return min3(x, y, z); } int fmin2(int x, int y, int z) { return min(min(x, y), z); } $ gcc -O2 -o min-max.s -S min-max.c; cat min-max.s .file "min-max.c" .text .p2align 4,,15 .globl fmin3 .type fmin3, @function fmin3: .LFB0: .cfi_startproc cmpl %esi, %edi jl .L5 cmpl %esi, %edx movl %esi, %eax cmovle %edx, %eax ret .p2align 4,,10 .p2align 3 .L5: cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE0: .size fmin3, .-fmin3 .p2align 4,,15 .globl fmin2 .type fmin2, @function fmin2: .LFB1: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovle %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE1: .size fmin2, .-fmin2 .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits fmin3 function, which uses open-coded min3 macro, is compiled into total of ten instructions including a conditional branch, whereas fmin2 function, which uses two calls to min2 macro, is compiled into six instructions with no branches. Similarly, open-coded clamp produces the same code as clamp using min and max macros, but the latter is much shorter: $ cat clamp.c #define clamp(val, min, max) ({ \ typeof(val) __val = (val); \ typeof(min) __min = (min); \ typeof(max) __max = (max); \ (void) (&__val == &__min); \ (void) (&__val == &__max); \ __val = __val < __min ? __min: __val; \ __val > __max ? __max: __val; }) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define max(x, y) ({ \ typeof(x) _max1 = (x); \ typeof(y) _max2 = (y); \ (void) (&_max1 == &_max2); \ _max1 > _max2 ? _max1 : _max2; }) int fclamp(int v, int min, int max) { return clamp(v, min, max); } int fclampmm(int v, int min, int max) { return min(max(v, min), max); } $ gcc -O2 -o clamp.s -S clamp.c; cat clamp.s .file "clamp.c" .text .p2align 4,,15 .globl fclamp .type fclamp, @function fclamp: .LFB0: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovge %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE0: .size fclamp, .-fclamp .p2align 4,,15 .globl fclampmm .type fclampmm, @function fclampmm: .LFB1: .cfi_startproc cmpl %edi, %esi cmovge %esi, %edi cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE1: .size fclampmm, .-fclampmm .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits Linux mpn-glaptop 3.13.0-29-generic #53~precise1-Ubuntu SMP Wed Jun 4 22:06:25 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 51224656 Jun 17 14:15 vmlinux.before -rwx------ 1 mpn eng 51224608 Jun 17 13:57 vmlinux.after 48 bytes reduction. The do_fault_around was a few instruction shorter and as far as I can tell saved 12 bytes on the stack, i.e.: $ grep -e rsp -e pop -e push do_fault_around.* do_fault_around.before.s:push %rbp do_fault_around.before.s:mov %rsp,%rbp do_fault_around.before.s:push %r13 do_fault_around.before.s:push %r12 do_fault_around.before.s:push %rbx do_fault_around.before.s:sub $0x38,%rsp do_fault_around.before.s:add $0x38,%rsp do_fault_around.before.s:pop %rbx do_fault_around.before.s:pop %r12 do_fault_around.before.s:pop %r13 do_fault_around.before.s:pop %rbp do_fault_around.after.s:push %rbp do_fault_around.after.s:mov %rsp,%rbp do_fault_around.after.s:push %r12 do_fault_around.after.s:push %rbx do_fault_around.after.s:sub $0x30,%rsp do_fault_around.after.s:add $0x30,%rsp do_fault_around.after.s:pop %rbx do_fault_around.after.s:pop %r12 do_fault_around.after.s:pop %rbp or here side-by-side: Before After push %rbp push %rbp mov %rsp,%rbp mov %rsp,%rbp push %r13 push %r12 push %r12 push %rbx push %rbx sub $0x38,%rsp sub $0x30,%rsp add $0x38,%rsp add $0x30,%rsp pop %rbx pop %rbx pop %r12 pop %r12 pop %r13 pop %rbp pop %rbp There are also fewer branches: $ grep ^j do_fault_around.* do_fault_around.before.s:jae ffffffff812079b7 do_fault_around.before.s:jmp ffffffff812079c5 do_fault_around.before.s:jmp ffffffff81207a14 do_fault_around.before.s:ja ffffffff812079f9 do_fault_around.before.s:jb ffffffff81207a10 do_fault_around.before.s:jmp ffffffff81207a63 do_fault_around.before.s:jne ffffffff812079df do_fault_around.after.s:jmp ffffffff812079fd do_fault_around.after.s:ja ffffffff812079e2 do_fault_around.after.s:jb ffffffff812079f9 do_fault_around.after.s:jmp ffffffff81207a4c do_fault_around.after.s:jne ffffffff812079c8 And here's with allyesconfig on a different machine: $ uname -a; gcc --version; ls -l vmlinux.* Linux erwin 3.14.7-mn torvalds#54 SMP Sun Jun 15 11:25:08 CEST 2014 x86_64 AMD Phenom(tm) II X3 710 Processor AuthenticAMD GNU/Linux gcc (GCC) 4.8.3 Copyright (C) 2013 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 437027411 Jun 20 16:04 vmlinux.before -rwx------ 1 mpn eng 437026881 Jun 20 15:30 vmlinux.after 530 bytes reduction. Signed-off-by: Michal Nazarewicz <[email protected]> Signed-off-by: Hagen Paul Pfeifer <[email protected]> Acked-by: Steven Rostedt <[email protected]> Cc: Hagen Paul Pfeifer <[email protected]> Cc: David Rientjes <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

@progbits

It appears that gcc is better at optimising a double call to min and max rather than open coded min3 and max3. This can be observed here: $ cat min-max.c #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define min3(x, y, z) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ typeof(z) _min3 = (z); \ (void) (&_min1 == &_min2); \ (void) (&_min1 == &_min3); \ _min1 < _min2 ? (_min1 < _min3 ? _min1 : _min3) : \ (_min2 < _min3 ? _min2 : _min3); }) int fmin3(int x, int y, int z) { return min3(x, y, z); } int fmin2(int x, int y, int z) { return min(min(x, y), z); } $ gcc -O2 -o min-max.s -S min-max.c; cat min-max.s .file "min-max.c" .text .p2align 4,,15 .globl fmin3 .type fmin3, @function fmin3: .LFB0: .cfi_startproc cmpl %esi, %edi jl .L5 cmpl %esi, %edx movl %esi, %eax cmovle %edx, %eax ret .p2align 4,,10 .p2align 3 .L5: cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE0: .size fmin3, .-fmin3 .p2align 4,,15 .globl fmin2 .type fmin2, @function fmin2: .LFB1: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovle %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE1: .size fmin2, .-fmin2 .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits fmin3 function, which uses open-coded min3 macro, is compiled into total of ten instructions including a conditional branch, whereas fmin2 function, which uses two calls to min2 macro, is compiled into six instructions with no branches. Similarly, open-coded clamp produces the same code as clamp using min and max macros, but the latter is much shorter: $ cat clamp.c #define clamp(val, min, max) ({ \ typeof(val) __val = (val); \ typeof(min) __min = (min); \ typeof(max) __max = (max); \ (void) (&__val == &__min); \ (void) (&__val == &__max); \ __val = __val < __min ? __min: __val; \ __val > __max ? __max: __val; }) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define max(x, y) ({ \ typeof(x) _max1 = (x); \ typeof(y) _max2 = (y); \ (void) (&_max1 == &_max2); \ _max1 > _max2 ? _max1 : _max2; }) int fclamp(int v, int min, int max) { return clamp(v, min, max); } int fclampmm(int v, int min, int max) { return min(max(v, min), max); } $ gcc -O2 -o clamp.s -S clamp.c; cat clamp.s .file "clamp.c" .text .p2align 4,,15 .globl fclamp .type fclamp, @function fclamp: .LFB0: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovge %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE0: .size fclamp, .-fclamp .p2align 4,,15 .globl fclampmm .type fclampmm, @function fclampmm: .LFB1: .cfi_startproc cmpl %edi, %esi cmovge %esi, %edi cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE1: .size fclampmm, .-fclampmm .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits Linux mpn-glaptop 3.13.0-29-generic #53~precise1-Ubuntu SMP Wed Jun 4 22:06:25 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 51224656 Jun 17 14:15 vmlinux.before -rwx------ 1 mpn eng 51224608 Jun 17 13:57 vmlinux.after 48 bytes reduction. The do_fault_around was a few instruction shorter and as far as I can tell saved 12 bytes on the stack, i.e.: $ grep -e rsp -e pop -e push do_fault_around.* do_fault_around.before.s:push %rbp do_fault_around.before.s:mov %rsp,%rbp do_fault_around.before.s:push %r13 do_fault_around.before.s:push %r12 do_fault_around.before.s:push %rbx do_fault_around.before.s:sub $0x38,%rsp do_fault_around.before.s:add $0x38,%rsp do_fault_around.before.s:pop %rbx do_fault_around.before.s:pop %r12 do_fault_around.before.s:pop %r13 do_fault_around.before.s:pop %rbp do_fault_around.after.s:push %rbp do_fault_around.after.s:mov %rsp,%rbp do_fault_around.after.s:push %r12 do_fault_around.after.s:push %rbx do_fault_around.after.s:sub $0x30,%rsp do_fault_around.after.s:add $0x30,%rsp do_fault_around.after.s:pop %rbx do_fault_around.after.s:pop %r12 do_fault_around.after.s:pop %rbp or here side-by-side: Before After push %rbp push %rbp mov %rsp,%rbp mov %rsp,%rbp push %r13 push %r12 push %r12 push %rbx push %rbx sub $0x38,%rsp sub $0x30,%rsp add $0x38,%rsp add $0x30,%rsp pop %rbx pop %rbx pop %r12 pop %r12 pop %r13 pop %rbp pop %rbp There are also fewer branches: $ grep ^j do_fault_around.* do_fault_around.before.s:jae ffffffff812079b7 do_fault_around.before.s:jmp ffffffff812079c5 do_fault_around.before.s:jmp ffffffff81207a14 do_fault_around.before.s:ja ffffffff812079f9 do_fault_around.before.s:jb ffffffff81207a10 do_fault_around.before.s:jmp ffffffff81207a63 do_fault_around.before.s:jne ffffffff812079df do_fault_around.after.s:jmp ffffffff812079fd do_fault_around.after.s:ja ffffffff812079e2 do_fault_around.after.s:jb ffffffff812079f9 do_fault_around.after.s:jmp ffffffff81207a4c do_fault_around.after.s:jne ffffffff812079c8 And here's with allyesconfig on a different machine: $ uname -a; gcc --version; ls -l vmlinux.* Linux erwin 3.14.7-mn torvalds#54 SMP Sun Jun 15 11:25:08 CEST 2014 x86_64 AMD Phenom(tm) II X3 710 Processor AuthenticAMD GNU/Linux gcc (GCC) 4.8.3 Copyright (C) 2013 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 437027411 Jun 20 16:04 vmlinux.before -rwx------ 1 mpn eng 437026881 Jun 20 15:30 vmlinux.after 530 bytes reduction. Signed-off-by: Michal Nazarewicz <[email protected]> Signed-off-by: Hagen Paul Pfeifer <[email protected]> Acked-by: Steven Rostedt <[email protected]> Cc: Hagen Paul Pfeifer <[email protected]> Cc: David Rientjes <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

@progbits

It appears that gcc is better at optimising a double call to min and max rather than open coded min3 and max3. This can be observed here: $ cat min-max.c #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define min3(x, y, z) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ typeof(z) _min3 = (z); \ (void) (&_min1 == &_min2); \ (void) (&_min1 == &_min3); \ _min1 < _min2 ? (_min1 < _min3 ? _min1 : _min3) : \ (_min2 < _min3 ? _min2 : _min3); }) int fmin3(int x, int y, int z) { return min3(x, y, z); } int fmin2(int x, int y, int z) { return min(min(x, y), z); } $ gcc -O2 -o min-max.s -S min-max.c; cat min-max.s .file "min-max.c" .text .p2align 4,,15 .globl fmin3 .type fmin3, @function fmin3: .LFB0: .cfi_startproc cmpl %esi, %edi jl .L5 cmpl %esi, %edx movl %esi, %eax cmovle %edx, %eax ret .p2align 4,,10 .p2align 3 .L5: cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE0: .size fmin3, .-fmin3 .p2align 4,,15 .globl fmin2 .type fmin2, @function fmin2: .LFB1: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovle %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE1: .size fmin2, .-fmin2 .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits fmin3 function, which uses open-coded min3 macro, is compiled into total of ten instructions including a conditional branch, whereas fmin2 function, which uses two calls to min2 macro, is compiled into six instructions with no branches. Similarly, open-coded clamp produces the same code as clamp using min and max macros, but the latter is much shorter: $ cat clamp.c #define clamp(val, min, max) ({ \ typeof(val) __val = (val); \ typeof(min) __min = (min); \ typeof(max) __max = (max); \ (void) (&__val == &__min); \ (void) (&__val == &__max); \ __val = __val < __min ? __min: __val; \ __val > __max ? __max: __val; }) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define max(x, y) ({ \ typeof(x) _max1 = (x); \ typeof(y) _max2 = (y); \ (void) (&_max1 == &_max2); \ _max1 > _max2 ? _max1 : _max2; }) int fclamp(int v, int min, int max) { return clamp(v, min, max); } int fclampmm(int v, int min, int max) { return min(max(v, min), max); } $ gcc -O2 -o clamp.s -S clamp.c; cat clamp.s .file "clamp.c" .text .p2align 4,,15 .globl fclamp .type fclamp, @function fclamp: .LFB0: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovge %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE0: .size fclamp, .-fclamp .p2align 4,,15 .globl fclampmm .type fclampmm, @function fclampmm: .LFB1: .cfi_startproc cmpl %edi, %esi cmovge %esi, %edi cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE1: .size fclampmm, .-fclampmm .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits Linux mpn-glaptop 3.13.0-29-generic #53~precise1-Ubuntu SMP Wed Jun 4 22:06:25 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 51224656 Jun 17 14:15 vmlinux.before -rwx------ 1 mpn eng 51224608 Jun 17 13:57 vmlinux.after 48 bytes reduction. The do_fault_around was a few instruction shorter and as far as I can tell saved 12 bytes on the stack, i.e.: $ grep -e rsp -e pop -e push do_fault_around.* do_fault_around.before.s:push %rbp do_fault_around.before.s:mov %rsp,%rbp do_fault_around.before.s:push %r13 do_fault_around.before.s:push %r12 do_fault_around.before.s:push %rbx do_fault_around.before.s:sub $0x38,%rsp do_fault_around.before.s:add $0x38,%rsp do_fault_around.before.s:pop %rbx do_fault_around.before.s:pop %r12 do_fault_around.before.s:pop %r13 do_fault_around.before.s:pop %rbp do_fault_around.after.s:push %rbp do_fault_around.after.s:mov %rsp,%rbp do_fault_around.after.s:push %r12 do_fault_around.after.s:push %rbx do_fault_around.after.s:sub $0x30,%rsp do_fault_around.after.s:add $0x30,%rsp do_fault_around.after.s:pop %rbx do_fault_around.after.s:pop %r12 do_fault_around.after.s:pop %rbp or here side-by-side: Before After push %rbp push %rbp mov %rsp,%rbp mov %rsp,%rbp push %r13 push %r12 push %r12 push %rbx push %rbx sub $0x38,%rsp sub $0x30,%rsp add $0x38,%rsp add $0x30,%rsp pop %rbx pop %rbx pop %r12 pop %r12 pop %r13 pop %rbp pop %rbp There are also fewer branches: $ grep ^j do_fault_around.* do_fault_around.before.s:jae ffffffff812079b7 do_fault_around.before.s:jmp ffffffff812079c5 do_fault_around.before.s:jmp ffffffff81207a14 do_fault_around.before.s:ja ffffffff812079f9 do_fault_around.before.s:jb ffffffff81207a10 do_fault_around.before.s:jmp ffffffff81207a63 do_fault_around.before.s:jne ffffffff812079df do_fault_around.after.s:jmp ffffffff812079fd do_fault_around.after.s:ja ffffffff812079e2 do_fault_around.after.s:jb ffffffff812079f9 do_fault_around.after.s:jmp ffffffff81207a4c do_fault_around.after.s:jne ffffffff812079c8 And here's with allyesconfig on a different machine: $ uname -a; gcc --version; ls -l vmlinux.* Linux erwin 3.14.7-mn torvalds#54 SMP Sun Jun 15 11:25:08 CEST 2014 x86_64 AMD Phenom(tm) II X3 710 Processor AuthenticAMD GNU/Linux gcc (GCC) 4.8.3 Copyright (C) 2013 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 437027411 Jun 20 16:04 vmlinux.before -rwx------ 1 mpn eng 437026881 Jun 20 15:30 vmlinux.after 530 bytes reduction. Signed-off-by: Michal Nazarewicz <[email protected]> Signed-off-by: Hagen Paul Pfeifer <[email protected]> Acked-by: Steven Rostedt <[email protected]> Cc: Hagen Paul Pfeifer <[email protected]> Cc: David Rientjes <[email protected]> Cc: "Rustad, Mark D" <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

Firmware-assisted dump (fadump) kernel code is not LE compliant. The below patch tries to fix this issue. Tested this patch with upstream kernel. Did some sanity testing for the LE fadump vmcore generated. Below output shows crash tool successfully opening LE fadump vmcore. # crash $vmlinux vmcore crash 7.0.5 Copyright (C) 2002-2014 Red Hat, Inc. Copyright (C) 2004, 2005, 2006, 2010 IBM Corporation Copyright (C) 1999-2006 Hewlett-Packard Co Copyright (C) 2005, 2006, 2011, 2012 Fujitsu Limited Copyright (C) 2006, 2007 VA Linux Systems Japan K.K. Copyright (C) 2005, 2011 NEC Corporation Copyright (C) 1999, 2002, 2007 Silicon Graphics, Inc. Copyright (C) 1999, 2000, 2001, 2002 Mission Critical Linux, Inc. This program is free software, covered by the GNU General Public License, and you are welcome to change it and/or distribute copies of it under certain conditions. Enter "help copying" to see the conditions. This program has absolutely no warranty. Enter "help warranty" for details. crash: /boot/vmlinux-3.16.0-rc7-7-default+: no .gnu_debuglink section GNU gdb (GDB) 7.6 Copyright (C) 2013 Free Software Foundation, Inc. License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html> This is free software: you are free to change and redistribute it. There is NO WARRANTY, to the extent permitted by law. Type "show copying" and "show warranty" for details. This GDB was configured as "powerpc64le-unknown-linux-gnu"... KERNEL: /boot/vmlinux-3.16.0-rc7-7-default+ DUMPFILE: vmcore CPUS: 16 DATE: Sun Aug 24 14:31:28 2014 UPTIME: 00:02:57 LOAD AVERAGE: 0.05, 0.08, 0.04 TASKS: 256 NODENAME: linux-dhr2 RELEASE: 3.16.0-rc7-7-default+ VERSION: torvalds#54 SMP Mon Aug 18 14:08:23 EDT 2014 MACHINE: ppc64le (4116 Mhz) MEMORY: 40 GB PANIC: "Oops: Kernel access of bad area, sig: 11 [#1]" (check log for details) PID: 2234 COMMAND: "bash" TASK: c0000009652e4a30 [THREAD_INFO: c00000096777c000] CPU: 2 STATE: TASK_RUNNING (PANIC) crash> Signed-off-by: Hari Bathini <[email protected]> Reviewed-by: Mahesh Salgaonkar <[email protected]>

@progbits

It appears that gcc is better at optimising a double call to min and max rather than open coded min3 and max3. This can be observed here: $ cat min-max.c #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define min3(x, y, z) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ typeof(z) _min3 = (z); \ (void) (&_min1 == &_min2); \ (void) (&_min1 == &_min3); \ _min1 < _min2 ? (_min1 < _min3 ? _min1 : _min3) : \ (_min2 < _min3 ? _min2 : _min3); }) int fmin3(int x, int y, int z) { return min3(x, y, z); } int fmin2(int x, int y, int z) { return min(min(x, y), z); } $ gcc -O2 -o min-max.s -S min-max.c; cat min-max.s .file "min-max.c" .text .p2align 4,,15 .globl fmin3 .type fmin3, @function fmin3: .LFB0: .cfi_startproc cmpl %esi, %edi jl .L5 cmpl %esi, %edx movl %esi, %eax cmovle %edx, %eax ret .p2align 4,,10 .p2align 3 .L5: cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE0: .size fmin3, .-fmin3 .p2align 4,,15 .globl fmin2 .type fmin2, @function fmin2: .LFB1: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovle %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE1: .size fmin2, .-fmin2 .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits fmin3 function, which uses open-coded min3 macro, is compiled into total of ten instructions including a conditional branch, whereas fmin2 function, which uses two calls to min2 macro, is compiled into six instructions with no branches. Similarly, open-coded clamp produces the same code as clamp using min and max macros, but the latter is much shorter: $ cat clamp.c #define clamp(val, min, max) ({ \ typeof(val) __val = (val); \ typeof(min) __min = (min); \ typeof(max) __max = (max); \ (void) (&__val == &__min); \ (void) (&__val == &__max); \ __val = __val < __min ? __min: __val; \ __val > __max ? __max: __val; }) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define max(x, y) ({ \ typeof(x) _max1 = (x); \ typeof(y) _max2 = (y); \ (void) (&_max1 == &_max2); \ _max1 > _max2 ? _max1 : _max2; }) int fclamp(int v, int min, int max) { return clamp(v, min, max); } int fclampmm(int v, int min, int max) { return min(max(v, min), max); } $ gcc -O2 -o clamp.s -S clamp.c; cat clamp.s .file "clamp.c" .text .p2align 4,,15 .globl fclamp .type fclamp, @function fclamp: .LFB0: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovge %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE0: .size fclamp, .-fclamp .p2align 4,,15 .globl fclampmm .type fclampmm, @function fclampmm: .LFB1: .cfi_startproc cmpl %edi, %esi cmovge %esi, %edi cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE1: .size fclampmm, .-fclampmm .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits Linux mpn-glaptop 3.13.0-29-generic #53~precise1-Ubuntu SMP Wed Jun 4 22:06:25 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 51224656 Jun 17 14:15 vmlinux.before -rwx------ 1 mpn eng 51224608 Jun 17 13:57 vmlinux.after 48 bytes reduction. The do_fault_around was a few instruction shorter and as far as I can tell saved 12 bytes on the stack, i.e.: $ grep -e rsp -e pop -e push do_fault_around.* do_fault_around.before.s:push %rbp do_fault_around.before.s:mov %rsp,%rbp do_fault_around.before.s:push %r13 do_fault_around.before.s:push %r12 do_fault_around.before.s:push %rbx do_fault_around.before.s:sub $0x38,%rsp do_fault_around.before.s:add $0x38,%rsp do_fault_around.before.s:pop %rbx do_fault_around.before.s:pop %r12 do_fault_around.before.s:pop %r13 do_fault_around.before.s:pop %rbp do_fault_around.after.s:push %rbp do_fault_around.after.s:mov %rsp,%rbp do_fault_around.after.s:push %r12 do_fault_around.after.s:push %rbx do_fault_around.after.s:sub $0x30,%rsp do_fault_around.after.s:add $0x30,%rsp do_fault_around.after.s:pop %rbx do_fault_around.after.s:pop %r12 do_fault_around.after.s:pop %rbp or here side-by-side: Before After push %rbp push %rbp mov %rsp,%rbp mov %rsp,%rbp push %r13 push %r12 push %r12 push %rbx push %rbx sub $0x38,%rsp sub $0x30,%rsp add $0x38,%rsp add $0x30,%rsp pop %rbx pop %rbx pop %r12 pop %r12 pop %r13 pop %rbp pop %rbp There are also fewer branches: $ grep ^j do_fault_around.* do_fault_around.before.s:jae ffffffff812079b7 do_fault_around.before.s:jmp ffffffff812079c5 do_fault_around.before.s:jmp ffffffff81207a14 do_fault_around.before.s:ja ffffffff812079f9 do_fault_around.before.s:jb ffffffff81207a10 do_fault_around.before.s:jmp ffffffff81207a63 do_fault_around.before.s:jne ffffffff812079df do_fault_around.after.s:jmp ffffffff812079fd do_fault_around.after.s:ja ffffffff812079e2 do_fault_around.after.s:jb ffffffff812079f9 do_fault_around.after.s:jmp ffffffff81207a4c do_fault_around.after.s:jne ffffffff812079c8 And here's with allyesconfig on a different machine: $ uname -a; gcc --version; ls -l vmlinux.* Linux erwin 3.14.7-mn torvalds#54 SMP Sun Jun 15 11:25:08 CEST 2014 x86_64 AMD Phenom(tm) II X3 710 Processor AuthenticAMD GNU/Linux gcc (GCC) 4.8.3 Copyright (C) 2013 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 437027411 Jun 20 16:04 vmlinux.before -rwx------ 1 mpn eng 437026881 Jun 20 15:30 vmlinux.after 530 bytes reduction. Signed-off-by: Michal Nazarewicz <[email protected]> Signed-off-by: Hagen Paul Pfeifer <[email protected]> Acked-by: Steven Rostedt <[email protected]> Cc: Hagen Paul Pfeifer <[email protected]> Cc: David Rientjes <[email protected]> Cc: "Rustad, Mark D" <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

@progbits

It appears that gcc is better at optimising a double call to min and max rather than open coded min3 and max3. This can be observed here: $ cat min-max.c #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define min3(x, y, z) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ typeof(z) _min3 = (z); \ (void) (&_min1 == &_min2); \ (void) (&_min1 == &_min3); \ _min1 < _min2 ? (_min1 < _min3 ? _min1 : _min3) : \ (_min2 < _min3 ? _min2 : _min3); }) int fmin3(int x, int y, int z) { return min3(x, y, z); } int fmin2(int x, int y, int z) { return min(min(x, y), z); } $ gcc -O2 -o min-max.s -S min-max.c; cat min-max.s .file "min-max.c" .text .p2align 4,,15 .globl fmin3 .type fmin3, @function fmin3: .LFB0: .cfi_startproc cmpl %esi, %edi jl .L5 cmpl %esi, %edx movl %esi, %eax cmovle %edx, %eax ret .p2align 4,,10 .p2align 3 .L5: cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE0: .size fmin3, .-fmin3 .p2align 4,,15 .globl fmin2 .type fmin2, @function fmin2: .LFB1: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovle %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE1: .size fmin2, .-fmin2 .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits fmin3 function, which uses open-coded min3 macro, is compiled into total of ten instructions including a conditional branch, whereas fmin2 function, which uses two calls to min2 macro, is compiled into six instructions with no branches. Similarly, open-coded clamp produces the same code as clamp using min and max macros, but the latter is much shorter: $ cat clamp.c #define clamp(val, min, max) ({ \ typeof(val) __val = (val); \ typeof(min) __min = (min); \ typeof(max) __max = (max); \ (void) (&__val == &__min); \ (void) (&__val == &__max); \ __val = __val < __min ? __min: __val; \ __val > __max ? __max: __val; }) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define max(x, y) ({ \ typeof(x) _max1 = (x); \ typeof(y) _max2 = (y); \ (void) (&_max1 == &_max2); \ _max1 > _max2 ? _max1 : _max2; }) int fclamp(int v, int min, int max) { return clamp(v, min, max); } int fclampmm(int v, int min, int max) { return min(max(v, min), max); } $ gcc -O2 -o clamp.s -S clamp.c; cat clamp.s .file "clamp.c" .text .p2align 4,,15 .globl fclamp .type fclamp, @function fclamp: .LFB0: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovge %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE0: .size fclamp, .-fclamp .p2align 4,,15 .globl fclampmm .type fclampmm, @function fclampmm: .LFB1: .cfi_startproc cmpl %edi, %esi cmovge %esi, %edi cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE1: .size fclampmm, .-fclampmm .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits Linux mpn-glaptop 3.13.0-29-generic #53~precise1-Ubuntu SMP Wed Jun 4 22:06:25 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 51224656 Jun 17 14:15 vmlinux.before -rwx------ 1 mpn eng 51224608 Jun 17 13:57 vmlinux.after 48 bytes reduction. The do_fault_around was a few instruction shorter and as far as I can tell saved 12 bytes on the stack, i.e.: $ grep -e rsp -e pop -e push do_fault_around.* do_fault_around.before.s:push %rbp do_fault_around.before.s:mov %rsp,%rbp do_fault_around.before.s:push %r13 do_fault_around.before.s:push %r12 do_fault_around.before.s:push %rbx do_fault_around.before.s:sub $0x38,%rsp do_fault_around.before.s:add $0x38,%rsp do_fault_around.before.s:pop %rbx do_fault_around.before.s:pop %r12 do_fault_around.before.s:pop %r13 do_fault_around.before.s:pop %rbp do_fault_around.after.s:push %rbp do_fault_around.after.s:mov %rsp,%rbp do_fault_around.after.s:push %r12 do_fault_around.after.s:push %rbx do_fault_around.after.s:sub $0x30,%rsp do_fault_around.after.s:add $0x30,%rsp do_fault_around.after.s:pop %rbx do_fault_around.after.s:pop %r12 do_fault_around.after.s:pop %rbp or here side-by-side: Before After push %rbp push %rbp mov %rsp,%rbp mov %rsp,%rbp push %r13 push %r12 push %r12 push %rbx push %rbx sub $0x38,%rsp sub $0x30,%rsp add $0x38,%rsp add $0x30,%rsp pop %rbx pop %rbx pop %r12 pop %r12 pop %r13 pop %rbp pop %rbp There are also fewer branches: $ grep ^j do_fault_around.* do_fault_around.before.s:jae ffffffff812079b7 do_fault_around.before.s:jmp ffffffff812079c5 do_fault_around.before.s:jmp ffffffff81207a14 do_fault_around.before.s:ja ffffffff812079f9 do_fault_around.before.s:jb ffffffff81207a10 do_fault_around.before.s:jmp ffffffff81207a63 do_fault_around.before.s:jne ffffffff812079df do_fault_around.after.s:jmp ffffffff812079fd do_fault_around.after.s:ja ffffffff812079e2 do_fault_around.after.s:jb ffffffff812079f9 do_fault_around.after.s:jmp ffffffff81207a4c do_fault_around.after.s:jne ffffffff812079c8 And here's with allyesconfig on a different machine: $ uname -a; gcc --version; ls -l vmlinux.* Linux erwin 3.14.7-mn torvalds#54 SMP Sun Jun 15 11:25:08 CEST 2014 x86_64 AMD Phenom(tm) II X3 710 Processor AuthenticAMD GNU/Linux gcc (GCC) 4.8.3 Copyright (C) 2013 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 437027411 Jun 20 16:04 vmlinux.before -rwx------ 1 mpn eng 437026881 Jun 20 15:30 vmlinux.after 530 bytes reduction. Signed-off-by: Michal Nazarewicz <[email protected]> Signed-off-by: Hagen Paul Pfeifer <[email protected]> Acked-by: Steven Rostedt <[email protected]> Cc: Hagen Paul Pfeifer <[email protected]> Cc: David Rientjes <[email protected]> Cc: "Rustad, Mark D" <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

@progbits

It appears that gcc is better at optimising a double call to min and max rather than open coded min3 and max3. This can be observed here: $ cat min-max.c #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define min3(x, y, z) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ typeof(z) _min3 = (z); \ (void) (&_min1 == &_min2); \ (void) (&_min1 == &_min3); \ _min1 < _min2 ? (_min1 < _min3 ? _min1 : _min3) : \ (_min2 < _min3 ? _min2 : _min3); }) int fmin3(int x, int y, int z) { return min3(x, y, z); } int fmin2(int x, int y, int z) { return min(min(x, y), z); } $ gcc -O2 -o min-max.s -S min-max.c; cat min-max.s .file "min-max.c" .text .p2align 4,,15 .globl fmin3 .type fmin3, @function fmin3: .LFB0: .cfi_startproc cmpl %esi, %edi jl .L5 cmpl %esi, %edx movl %esi, %eax cmovle %edx, %eax ret .p2align 4,,10 .p2align 3 .L5: cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE0: .size fmin3, .-fmin3 .p2align 4,,15 .globl fmin2 .type fmin2, @function fmin2: .LFB1: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovle %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE1: .size fmin2, .-fmin2 .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits fmin3 function, which uses open-coded min3 macro, is compiled into total of ten instructions including a conditional branch, whereas fmin2 function, which uses two calls to min2 macro, is compiled into six instructions with no branches. Similarly, open-coded clamp produces the same code as clamp using min and max macros, but the latter is much shorter: $ cat clamp.c #define clamp(val, min, max) ({ \ typeof(val) __val = (val); \ typeof(min) __min = (min); \ typeof(max) __max = (max); \ (void) (&__val == &__min); \ (void) (&__val == &__max); \ __val = __val < __min ? __min: __val; \ __val > __max ? __max: __val; }) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define max(x, y) ({ \ typeof(x) _max1 = (x); \ typeof(y) _max2 = (y); \ (void) (&_max1 == &_max2); \ _max1 > _max2 ? _max1 : _max2; }) int fclamp(int v, int min, int max) { return clamp(v, min, max); } int fclampmm(int v, int min, int max) { return min(max(v, min), max); } $ gcc -O2 -o clamp.s -S clamp.c; cat clamp.s .file "clamp.c" .text .p2align 4,,15 .globl fclamp .type fclamp, @function fclamp: .LFB0: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovge %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE0: .size fclamp, .-fclamp .p2align 4,,15 .globl fclampmm .type fclampmm, @function fclampmm: .LFB1: .cfi_startproc cmpl %edi, %esi cmovge %esi, %edi cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE1: .size fclampmm, .-fclampmm .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits Linux mpn-glaptop 3.13.0-29-generic #53~precise1-Ubuntu SMP Wed Jun 4 22:06:25 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 51224656 Jun 17 14:15 vmlinux.before -rwx------ 1 mpn eng 51224608 Jun 17 13:57 vmlinux.after 48 bytes reduction. The do_fault_around was a few instruction shorter and as far as I can tell saved 12 bytes on the stack, i.e.: $ grep -e rsp -e pop -e push do_fault_around.* do_fault_around.before.s:push %rbp do_fault_around.before.s:mov %rsp,%rbp do_fault_around.before.s:push %r13 do_fault_around.before.s:push %r12 do_fault_around.before.s:push %rbx do_fault_around.before.s:sub $0x38,%rsp do_fault_around.before.s:add $0x38,%rsp do_fault_around.before.s:pop %rbx do_fault_around.before.s:pop %r12 do_fault_around.before.s:pop %r13 do_fault_around.before.s:pop %rbp do_fault_around.after.s:push %rbp do_fault_around.after.s:mov %rsp,%rbp do_fault_around.after.s:push %r12 do_fault_around.after.s:push %rbx do_fault_around.after.s:sub $0x30,%rsp do_fault_around.after.s:add $0x30,%rsp do_fault_around.after.s:pop %rbx do_fault_around.after.s:pop %r12 do_fault_around.after.s:pop %rbp or here side-by-side: Before After push %rbp push %rbp mov %rsp,%rbp mov %rsp,%rbp push %r13 push %r12 push %r12 push %rbx push %rbx sub $0x38,%rsp sub $0x30,%rsp add $0x38,%rsp add $0x30,%rsp pop %rbx pop %rbx pop %r12 pop %r12 pop %r13 pop %rbp pop %rbp There are also fewer branches: $ grep ^j do_fault_around.* do_fault_around.before.s:jae ffffffff812079b7 do_fault_around.before.s:jmp ffffffff812079c5 do_fault_around.before.s:jmp ffffffff81207a14 do_fault_around.before.s:ja ffffffff812079f9 do_fault_around.before.s:jb ffffffff81207a10 do_fault_around.before.s:jmp ffffffff81207a63 do_fault_around.before.s:jne ffffffff812079df do_fault_around.after.s:jmp ffffffff812079fd do_fault_around.after.s:ja ffffffff812079e2 do_fault_around.after.s:jb ffffffff812079f9 do_fault_around.after.s:jmp ffffffff81207a4c do_fault_around.after.s:jne ffffffff812079c8 And here's with allyesconfig on a different machine: $ uname -a; gcc --version; ls -l vmlinux.* Linux erwin 3.14.7-mn torvalds#54 SMP Sun Jun 15 11:25:08 CEST 2014 x86_64 AMD Phenom(tm) II X3 710 Processor AuthenticAMD GNU/Linux gcc (GCC) 4.8.3 Copyright (C) 2013 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 437027411 Jun 20 16:04 vmlinux.before -rwx------ 1 mpn eng 437026881 Jun 20 15:30 vmlinux.after 530 bytes reduction. Signed-off-by: Michal Nazarewicz <[email protected]> Signed-off-by: Hagen Paul Pfeifer <[email protected]> Acked-by: Steven Rostedt <[email protected]> Cc: Hagen Paul Pfeifer <[email protected]> Cc: David Rientjes <[email protected]> Cc: "Rustad, Mark D" <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

@progbits

It appears that gcc is better at optimising a double call to min and max rather than open coded min3 and max3. This can be observed here: $ cat min-max.c #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define min3(x, y, z) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ typeof(z) _min3 = (z); \ (void) (&_min1 == &_min2); \ (void) (&_min1 == &_min3); \ _min1 < _min2 ? (_min1 < _min3 ? _min1 : _min3) : \ (_min2 < _min3 ? _min2 : _min3); }) int fmin3(int x, int y, int z) { return min3(x, y, z); } int fmin2(int x, int y, int z) { return min(min(x, y), z); } $ gcc -O2 -o min-max.s -S min-max.c; cat min-max.s .file "min-max.c" .text .p2align 4,,15 .globl fmin3 .type fmin3, @function fmin3: .LFB0: .cfi_startproc cmpl %esi, %edi jl .L5 cmpl %esi, %edx movl %esi, %eax cmovle %edx, %eax ret .p2align 4,,10 .p2align 3 .L5: cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE0: .size fmin3, .-fmin3 .p2align 4,,15 .globl fmin2 .type fmin2, @function fmin2: .LFB1: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovle %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE1: .size fmin2, .-fmin2 .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits fmin3 function, which uses open-coded min3 macro, is compiled into total of ten instructions including a conditional branch, whereas fmin2 function, which uses two calls to min2 macro, is compiled into six instructions with no branches. Similarly, open-coded clamp produces the same code as clamp using min and max macros, but the latter is much shorter: $ cat clamp.c #define clamp(val, min, max) ({ \ typeof(val) __val = (val); \ typeof(min) __min = (min); \ typeof(max) __max = (max); \ (void) (&__val == &__min); \ (void) (&__val == &__max); \ __val = __val < __min ? __min: __val; \ __val > __max ? __max: __val; }) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define max(x, y) ({ \ typeof(x) _max1 = (x); \ typeof(y) _max2 = (y); \ (void) (&_max1 == &_max2); \ _max1 > _max2 ? _max1 : _max2; }) int fclamp(int v, int min, int max) { return clamp(v, min, max); } int fclampmm(int v, int min, int max) { return min(max(v, min), max); } $ gcc -O2 -o clamp.s -S clamp.c; cat clamp.s .file "clamp.c" .text .p2align 4,,15 .globl fclamp .type fclamp, @function fclamp: .LFB0: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovge %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE0: .size fclamp, .-fclamp .p2align 4,,15 .globl fclampmm .type fclampmm, @function fclampmm: .LFB1: .cfi_startproc cmpl %edi, %esi cmovge %esi, %edi cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE1: .size fclampmm, .-fclampmm .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits Linux mpn-glaptop 3.13.0-29-generic #53~precise1-Ubuntu SMP Wed Jun 4 22:06:25 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 51224656 Jun 17 14:15 vmlinux.before -rwx------ 1 mpn eng 51224608 Jun 17 13:57 vmlinux.after 48 bytes reduction. The do_fault_around was a few instruction shorter and as far as I can tell saved 12 bytes on the stack, i.e.: $ grep -e rsp -e pop -e push do_fault_around.* do_fault_around.before.s:push %rbp do_fault_around.before.s:mov %rsp,%rbp do_fault_around.before.s:push %r13 do_fault_around.before.s:push %r12 do_fault_around.before.s:push %rbx do_fault_around.before.s:sub $0x38,%rsp do_fault_around.before.s:add $0x38,%rsp do_fault_around.before.s:pop %rbx do_fault_around.before.s:pop %r12 do_fault_around.before.s:pop %r13 do_fault_around.before.s:pop %rbp do_fault_around.after.s:push %rbp do_fault_around.after.s:mov %rsp,%rbp do_fault_around.after.s:push %r12 do_fault_around.after.s:push %rbx do_fault_around.after.s:sub $0x30,%rsp do_fault_around.after.s:add $0x30,%rsp do_fault_around.after.s:pop %rbx do_fault_around.after.s:pop %r12 do_fault_around.after.s:pop %rbp or here side-by-side: Before After push %rbp push %rbp mov %rsp,%rbp mov %rsp,%rbp push %r13 push %r12 push %r12 push %rbx push %rbx sub $0x38,%rsp sub $0x30,%rsp add $0x38,%rsp add $0x30,%rsp pop %rbx pop %rbx pop %r12 pop %r12 pop %r13 pop %rbp pop %rbp There are also fewer branches: $ grep ^j do_fault_around.* do_fault_around.before.s:jae ffffffff812079b7 do_fault_around.before.s:jmp ffffffff812079c5 do_fault_around.before.s:jmp ffffffff81207a14 do_fault_around.before.s:ja ffffffff812079f9 do_fault_around.before.s:jb ffffffff81207a10 do_fault_around.before.s:jmp ffffffff81207a63 do_fault_around.before.s:jne ffffffff812079df do_fault_around.after.s:jmp ffffffff812079fd do_fault_around.after.s:ja ffffffff812079e2 do_fault_around.after.s:jb ffffffff812079f9 do_fault_around.after.s:jmp ffffffff81207a4c do_fault_around.after.s:jne ffffffff812079c8 And here's with allyesconfig on a different machine: $ uname -a; gcc --version; ls -l vmlinux.* Linux erwin 3.14.7-mn torvalds#54 SMP Sun Jun 15 11:25:08 CEST 2014 x86_64 AMD Phenom(tm) II X3 710 Processor AuthenticAMD GNU/Linux gcc (GCC) 4.8.3 Copyright (C) 2013 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 437027411 Jun 20 16:04 vmlinux.before -rwx------ 1 mpn eng 437026881 Jun 20 15:30 vmlinux.after 530 bytes reduction. Signed-off-by: Michal Nazarewicz <[email protected]> Signed-off-by: Hagen Paul Pfeifer <[email protected]> Acked-by: Steven Rostedt <[email protected]> Cc: Hagen Paul Pfeifer <[email protected]> Cc: David Rientjes <[email protected]> Cc: "Rustad, Mark D" <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

@progbits

It appears that gcc is better at optimising a double call to min and max rather than open coded min3 and max3. This can be observed here: $ cat min-max.c #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define min3(x, y, z) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ typeof(z) _min3 = (z); \ (void) (&_min1 == &_min2); \ (void) (&_min1 == &_min3); \ _min1 < _min2 ? (_min1 < _min3 ? _min1 : _min3) : \ (_min2 < _min3 ? _min2 : _min3); }) int fmin3(int x, int y, int z) { return min3(x, y, z); } int fmin2(int x, int y, int z) { return min(min(x, y), z); } $ gcc -O2 -o min-max.s -S min-max.c; cat min-max.s .file "min-max.c" .text .p2align 4,,15 .globl fmin3 .type fmin3, @function fmin3: .LFB0: .cfi_startproc cmpl %esi, %edi jl .L5 cmpl %esi, %edx movl %esi, %eax cmovle %edx, %eax ret .p2align 4,,10 .p2align 3 .L5: cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE0: .size fmin3, .-fmin3 .p2align 4,,15 .globl fmin2 .type fmin2, @function fmin2: .LFB1: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovle %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE1: .size fmin2, .-fmin2 .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits fmin3 function, which uses open-coded min3 macro, is compiled into total of ten instructions including a conditional branch, whereas fmin2 function, which uses two calls to min2 macro, is compiled into six instructions with no branches. Similarly, open-coded clamp produces the same code as clamp using min and max macros, but the latter is much shorter: $ cat clamp.c #define clamp(val, min, max) ({ \ typeof(val) __val = (val); \ typeof(min) __min = (min); \ typeof(max) __max = (max); \ (void) (&__val == &__min); \ (void) (&__val == &__max); \ __val = __val < __min ? __min: __val; \ __val > __max ? __max: __val; }) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define max(x, y) ({ \ typeof(x) _max1 = (x); \ typeof(y) _max2 = (y); \ (void) (&_max1 == &_max2); \ _max1 > _max2 ? _max1 : _max2; }) int fclamp(int v, int min, int max) { return clamp(v, min, max); } int fclampmm(int v, int min, int max) { return min(max(v, min), max); } $ gcc -O2 -o clamp.s -S clamp.c; cat clamp.s .file "clamp.c" .text .p2align 4,,15 .globl fclamp .type fclamp, @function fclamp: .LFB0: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovge %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE0: .size fclamp, .-fclamp .p2align 4,,15 .globl fclampmm .type fclampmm, @function fclampmm: .LFB1: .cfi_startproc cmpl %edi, %esi cmovge %esi, %edi cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE1: .size fclampmm, .-fclampmm .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits Linux mpn-glaptop 3.13.0-29-generic #53~precise1-Ubuntu SMP Wed Jun 4 22:06:25 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 51224656 Jun 17 14:15 vmlinux.before -rwx------ 1 mpn eng 51224608 Jun 17 13:57 vmlinux.after 48 bytes reduction. The do_fault_around was a few instruction shorter and as far as I can tell saved 12 bytes on the stack, i.e.: $ grep -e rsp -e pop -e push do_fault_around.* do_fault_around.before.s:push %rbp do_fault_around.before.s:mov %rsp,%rbp do_fault_around.before.s:push %r13 do_fault_around.before.s:push %r12 do_fault_around.before.s:push %rbx do_fault_around.before.s:sub $0x38,%rsp do_fault_around.before.s:add $0x38,%rsp do_fault_around.before.s:pop %rbx do_fault_around.before.s:pop %r12 do_fault_around.before.s:pop %r13 do_fault_around.before.s:pop %rbp do_fault_around.after.s:push %rbp do_fault_around.after.s:mov %rsp,%rbp do_fault_around.after.s:push %r12 do_fault_around.after.s:push %rbx do_fault_around.after.s:sub $0x30,%rsp do_fault_around.after.s:add $0x30,%rsp do_fault_around.after.s:pop %rbx do_fault_around.after.s:pop %r12 do_fault_around.after.s:pop %rbp or here side-by-side: Before After push %rbp push %rbp mov %rsp,%rbp mov %rsp,%rbp push %r13 push %r12 push %r12 push %rbx push %rbx sub $0x38,%rsp sub $0x30,%rsp add $0x38,%rsp add $0x30,%rsp pop %rbx pop %rbx pop %r12 pop %r12 pop %r13 pop %rbp pop %rbp There are also fewer branches: $ grep ^j do_fault_around.* do_fault_around.before.s:jae ffffffff812079b7 do_fault_around.before.s:jmp ffffffff812079c5 do_fault_around.before.s:jmp ffffffff81207a14 do_fault_around.before.s:ja ffffffff812079f9 do_fault_around.before.s:jb ffffffff81207a10 do_fault_around.before.s:jmp ffffffff81207a63 do_fault_around.before.s:jne ffffffff812079df do_fault_around.after.s:jmp ffffffff812079fd do_fault_around.after.s:ja ffffffff812079e2 do_fault_around.after.s:jb ffffffff812079f9 do_fault_around.after.s:jmp ffffffff81207a4c do_fault_around.after.s:jne ffffffff812079c8 And here's with allyesconfig on a different machine: $ uname -a; gcc --version; ls -l vmlinux.* Linux erwin 3.14.7-mn torvalds#54 SMP Sun Jun 15 11:25:08 CEST 2014 x86_64 AMD Phenom(tm) II X3 710 Processor AuthenticAMD GNU/Linux gcc (GCC) 4.8.3 Copyright (C) 2013 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 437027411 Jun 20 16:04 vmlinux.before -rwx------ 1 mpn eng 437026881 Jun 20 15:30 vmlinux.after 530 bytes reduction. Signed-off-by: Michal Nazarewicz <[email protected]> Signed-off-by: Hagen Paul Pfeifer <[email protected]> Acked-by: Steven Rostedt <[email protected]> Cc: Hagen Paul Pfeifer <[email protected]> Cc: David Rientjes <[email protected]> Cc: "Rustad, Mark D" <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

@progbits

It appears that gcc is better at optimising a double call to min and max rather than open coded min3 and max3. This can be observed here: $ cat min-max.c #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define min3(x, y, z) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ typeof(z) _min3 = (z); \ (void) (&_min1 == &_min2); \ (void) (&_min1 == &_min3); \ _min1 < _min2 ? (_min1 < _min3 ? _min1 : _min3) : \ (_min2 < _min3 ? _min2 : _min3); }) int fmin3(int x, int y, int z) { return min3(x, y, z); } int fmin2(int x, int y, int z) { return min(min(x, y), z); } $ gcc -O2 -o min-max.s -S min-max.c; cat min-max.s .file "min-max.c" .text .p2align 4,,15 .globl fmin3 .type fmin3, @function fmin3: .LFB0: .cfi_startproc cmpl %esi, %edi jl .L5 cmpl %esi, %edx movl %esi, %eax cmovle %edx, %eax ret .p2align 4,,10 .p2align 3 .L5: cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE0: .size fmin3, .-fmin3 .p2align 4,,15 .globl fmin2 .type fmin2, @function fmin2: .LFB1: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovle %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE1: .size fmin2, .-fmin2 .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits fmin3 function, which uses open-coded min3 macro, is compiled into total of ten instructions including a conditional branch, whereas fmin2 function, which uses two calls to min2 macro, is compiled into six instructions with no branches. Similarly, open-coded clamp produces the same code as clamp using min and max macros, but the latter is much shorter: $ cat clamp.c #define clamp(val, min, max) ({ \ typeof(val) __val = (val); \ typeof(min) __min = (min); \ typeof(max) __max = (max); \ (void) (&__val == &__min); \ (void) (&__val == &__max); \ __val = __val < __min ? __min: __val; \ __val > __max ? __max: __val; }) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define max(x, y) ({ \ typeof(x) _max1 = (x); \ typeof(y) _max2 = (y); \ (void) (&_max1 == &_max2); \ _max1 > _max2 ? _max1 : _max2; }) int fclamp(int v, int min, int max) { return clamp(v, min, max); } int fclampmm(int v, int min, int max) { return min(max(v, min), max); } $ gcc -O2 -o clamp.s -S clamp.c; cat clamp.s .file "clamp.c" .text .p2align 4,,15 .globl fclamp .type fclamp, @function fclamp: .LFB0: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovge %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE0: .size fclamp, .-fclamp .p2align 4,,15 .globl fclampmm .type fclampmm, @function fclampmm: .LFB1: .cfi_startproc cmpl %edi, %esi cmovge %esi, %edi cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE1: .size fclampmm, .-fclampmm .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits Linux mpn-glaptop 3.13.0-29-generic #53~precise1-Ubuntu SMP Wed Jun 4 22:06:25 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 51224656 Jun 17 14:15 vmlinux.before -rwx------ 1 mpn eng 51224608 Jun 17 13:57 vmlinux.after 48 bytes reduction. The do_fault_around was a few instruction shorter and as far as I can tell saved 12 bytes on the stack, i.e.: $ grep -e rsp -e pop -e push do_fault_around.* do_fault_around.before.s:push %rbp do_fault_around.before.s:mov %rsp,%rbp do_fault_around.before.s:push %r13 do_fault_around.before.s:push %r12 do_fault_around.before.s:push %rbx do_fault_around.before.s:sub $0x38,%rsp do_fault_around.before.s:add $0x38,%rsp do_fault_around.before.s:pop %rbx do_fault_around.before.s:pop %r12 do_fault_around.before.s:pop %r13 do_fault_around.before.s:pop %rbp do_fault_around.after.s:push %rbp do_fault_around.after.s:mov %rsp,%rbp do_fault_around.after.s:push %r12 do_fault_around.after.s:push %rbx do_fault_around.after.s:sub $0x30,%rsp do_fault_around.after.s:add $0x30,%rsp do_fault_around.after.s:pop %rbx do_fault_around.after.s:pop %r12 do_fault_around.after.s:pop %rbp or here side-by-side: Before After push %rbp push %rbp mov %rsp,%rbp mov %rsp,%rbp push %r13 push %r12 push %r12 push %rbx push %rbx sub $0x38,%rsp sub $0x30,%rsp add $0x38,%rsp add $0x30,%rsp pop %rbx pop %rbx pop %r12 pop %r12 pop %r13 pop %rbp pop %rbp There are also fewer branches: $ grep ^j do_fault_around.* do_fault_around.before.s:jae ffffffff812079b7 do_fault_around.before.s:jmp ffffffff812079c5 do_fault_around.before.s:jmp ffffffff81207a14 do_fault_around.before.s:ja ffffffff812079f9 do_fault_around.before.s:jb ffffffff81207a10 do_fault_around.before.s:jmp ffffffff81207a63 do_fault_around.before.s:jne ffffffff812079df do_fault_around.after.s:jmp ffffffff812079fd do_fault_around.after.s:ja ffffffff812079e2 do_fault_around.after.s:jb ffffffff812079f9 do_fault_around.after.s:jmp ffffffff81207a4c do_fault_around.after.s:jne ffffffff812079c8 And here's with allyesconfig on a different machine: $ uname -a; gcc --version; ls -l vmlinux.* Linux erwin 3.14.7-mn torvalds#54 SMP Sun Jun 15 11:25:08 CEST 2014 x86_64 AMD Phenom(tm) II X3 710 Processor AuthenticAMD GNU/Linux gcc (GCC) 4.8.3 Copyright (C) 2013 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 437027411 Jun 20 16:04 vmlinux.before -rwx------ 1 mpn eng 437026881 Jun 20 15:30 vmlinux.after 530 bytes reduction. Signed-off-by: Michal Nazarewicz <[email protected]> Signed-off-by: Hagen Paul Pfeifer <[email protected]> Acked-by: Steven Rostedt <[email protected]> Cc: Hagen Paul Pfeifer <[email protected]> Cc: David Rientjes <[email protected]> Cc: "Rustad, Mark D" <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

@progbits

It appears that gcc is better at optimising a double call to min and max rather than open coded min3 and max3. This can be observed here: $ cat min-max.c #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define min3(x, y, z) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ typeof(z) _min3 = (z); \ (void) (&_min1 == &_min2); \ (void) (&_min1 == &_min3); \ _min1 < _min2 ? (_min1 < _min3 ? _min1 : _min3) : \ (_min2 < _min3 ? _min2 : _min3); }) int fmin3(int x, int y, int z) { return min3(x, y, z); } int fmin2(int x, int y, int z) { return min(min(x, y), z); } $ gcc -O2 -o min-max.s -S min-max.c; cat min-max.s .file "min-max.c" .text .p2align 4,,15 .globl fmin3 .type fmin3, @function fmin3: .LFB0: .cfi_startproc cmpl %esi, %edi jl .L5 cmpl %esi, %edx movl %esi, %eax cmovle %edx, %eax ret .p2align 4,,10 .p2align 3 .L5: cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE0: .size fmin3, .-fmin3 .p2align 4,,15 .globl fmin2 .type fmin2, @function fmin2: .LFB1: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovle %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE1: .size fmin2, .-fmin2 .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits fmin3 function, which uses open-coded min3 macro, is compiled into total of ten instructions including a conditional branch, whereas fmin2 function, which uses two calls to min2 macro, is compiled into six instructions with no branches. Similarly, open-coded clamp produces the same code as clamp using min and max macros, but the latter is much shorter: $ cat clamp.c #define clamp(val, min, max) ({ \ typeof(val) __val = (val); \ typeof(min) __min = (min); \ typeof(max) __max = (max); \ (void) (&__val == &__min); \ (void) (&__val == &__max); \ __val = __val < __min ? __min: __val; \ __val > __max ? __max: __val; }) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define max(x, y) ({ \ typeof(x) _max1 = (x); \ typeof(y) _max2 = (y); \ (void) (&_max1 == &_max2); \ _max1 > _max2 ? _max1 : _max2; }) int fclamp(int v, int min, int max) { return clamp(v, min, max); } int fclampmm(int v, int min, int max) { return min(max(v, min), max); } $ gcc -O2 -o clamp.s -S clamp.c; cat clamp.s .file "clamp.c" .text .p2align 4,,15 .globl fclamp .type fclamp, @function fclamp: .LFB0: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovge %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE0: .size fclamp, .-fclamp .p2align 4,,15 .globl fclampmm .type fclampmm, @function fclampmm: .LFB1: .cfi_startproc cmpl %edi, %esi cmovge %esi, %edi cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE1: .size fclampmm, .-fclampmm .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits Linux mpn-glaptop 3.13.0-29-generic #53~precise1-Ubuntu SMP Wed Jun 4 22:06:25 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 51224656 Jun 17 14:15 vmlinux.before -rwx------ 1 mpn eng 51224608 Jun 17 13:57 vmlinux.after 48 bytes reduction. The do_fault_around was a few instruction shorter and as far as I can tell saved 12 bytes on the stack, i.e.: $ grep -e rsp -e pop -e push do_fault_around.* do_fault_around.before.s:push %rbp do_fault_around.before.s:mov %rsp,%rbp do_fault_around.before.s:push %r13 do_fault_around.before.s:push %r12 do_fault_around.before.s:push %rbx do_fault_around.before.s:sub $0x38,%rsp do_fault_around.before.s:add $0x38,%rsp do_fault_around.before.s:pop %rbx do_fault_around.before.s:pop %r12 do_fault_around.before.s:pop %r13 do_fault_around.before.s:pop %rbp do_fault_around.after.s:push %rbp do_fault_around.after.s:mov %rsp,%rbp do_fault_around.after.s:push %r12 do_fault_around.after.s:push %rbx do_fault_around.after.s:sub $0x30,%rsp do_fault_around.after.s:add $0x30,%rsp do_fault_around.after.s:pop %rbx do_fault_around.after.s:pop %r12 do_fault_around.after.s:pop %rbp or here side-by-side: Before After push %rbp push %rbp mov %rsp,%rbp mov %rsp,%rbp push %r13 push %r12 push %r12 push %rbx push %rbx sub $0x38,%rsp sub $0x30,%rsp add $0x38,%rsp add $0x30,%rsp pop %rbx pop %rbx pop %r12 pop %r12 pop %r13 pop %rbp pop %rbp There are also fewer branches: $ grep ^j do_fault_around.* do_fault_around.before.s:jae ffffffff812079b7 do_fault_around.before.s:jmp ffffffff812079c5 do_fault_around.before.s:jmp ffffffff81207a14 do_fault_around.before.s:ja ffffffff812079f9 do_fault_around.before.s:jb ffffffff81207a10 do_fault_around.before.s:jmp ffffffff81207a63 do_fault_around.before.s:jne ffffffff812079df do_fault_around.after.s:jmp ffffffff812079fd do_fault_around.after.s:ja ffffffff812079e2 do_fault_around.after.s:jb ffffffff812079f9 do_fault_around.after.s:jmp ffffffff81207a4c do_fault_around.after.s:jne ffffffff812079c8 And here's with allyesconfig on a different machine: $ uname -a; gcc --version; ls -l vmlinux.* Linux erwin 3.14.7-mn torvalds#54 SMP Sun Jun 15 11:25:08 CEST 2014 x86_64 AMD Phenom(tm) II X3 710 Processor AuthenticAMD GNU/Linux gcc (GCC) 4.8.3 Copyright (C) 2013 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 437027411 Jun 20 16:04 vmlinux.before -rwx------ 1 mpn eng 437026881 Jun 20 15:30 vmlinux.after 530 bytes reduction. Signed-off-by: Michal Nazarewicz <[email protected]> Signed-off-by: Hagen Paul Pfeifer <[email protected]> Acked-by: Steven Rostedt <[email protected]> Cc: Hagen Paul Pfeifer <[email protected]> Cc: David Rientjes <[email protected]> Cc: "Rustad, Mark D" <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

@progbits

It appears that gcc is better at optimising a double call to min and max rather than open coded min3 and max3. This can be observed here: $ cat min-max.c #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define min3(x, y, z) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ typeof(z) _min3 = (z); \ (void) (&_min1 == &_min2); \ (void) (&_min1 == &_min3); \ _min1 < _min2 ? (_min1 < _min3 ? _min1 : _min3) : \ (_min2 < _min3 ? _min2 : _min3); }) int fmin3(int x, int y, int z) { return min3(x, y, z); } int fmin2(int x, int y, int z) { return min(min(x, y), z); } $ gcc -O2 -o min-max.s -S min-max.c; cat min-max.s .file "min-max.c" .text .p2align 4,,15 .globl fmin3 .type fmin3, @function fmin3: .LFB0: .cfi_startproc cmpl %esi, %edi jl .L5 cmpl %esi, %edx movl %esi, %eax cmovle %edx, %eax ret .p2align 4,,10 .p2align 3 .L5: cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE0: .size fmin3, .-fmin3 .p2align 4,,15 .globl fmin2 .type fmin2, @function fmin2: .LFB1: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovle %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE1: .size fmin2, .-fmin2 .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits fmin3 function, which uses open-coded min3 macro, is compiled into total of ten instructions including a conditional branch, whereas fmin2 function, which uses two calls to min2 macro, is compiled into six instructions with no branches. Similarly, open-coded clamp produces the same code as clamp using min and max macros, but the latter is much shorter: $ cat clamp.c #define clamp(val, min, max) ({ \ typeof(val) __val = (val); \ typeof(min) __min = (min); \ typeof(max) __max = (max); \ (void) (&__val == &__min); \ (void) (&__val == &__max); \ __val = __val < __min ? __min: __val; \ __val > __max ? __max: __val; }) #define min(x, y) ({ \ typeof(x) _min1 = (x); \ typeof(y) _min2 = (y); \ (void) (&_min1 == &_min2); \ _min1 < _min2 ? _min1 : _min2; }) #define max(x, y) ({ \ typeof(x) _max1 = (x); \ typeof(y) _max2 = (y); \ (void) (&_max1 == &_max2); \ _max1 > _max2 ? _max1 : _max2; }) int fclamp(int v, int min, int max) { return clamp(v, min, max); } int fclampmm(int v, int min, int max) { return min(max(v, min), max); } $ gcc -O2 -o clamp.s -S clamp.c; cat clamp.s .file "clamp.c" .text .p2align 4,,15 .globl fclamp .type fclamp, @function fclamp: .LFB0: .cfi_startproc cmpl %edi, %esi movl %edx, %eax cmovge %esi, %edi cmpl %edx, %edi cmovle %edi, %eax ret .cfi_endproc .LFE0: .size fclamp, .-fclamp .p2align 4,,15 .globl fclampmm .type fclampmm, @function fclampmm: .LFB1: .cfi_startproc cmpl %edi, %esi cmovge %esi, %edi cmpl %edi, %edx movl %edi, %eax cmovle %edx, %eax ret .cfi_endproc .LFE1: .size fclampmm, .-fclampmm .ident "GCC: (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3" .section .note.GNU-stack,"",@progbits Linux mpn-glaptop 3.13.0-29-generic #53~precise1-Ubuntu SMP Wed Jun 4 22:06:25 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 Copyright (C) 2011 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 51224656 Jun 17 14:15 vmlinux.before -rwx------ 1 mpn eng 51224608 Jun 17 13:57 vmlinux.after 48 bytes reduction. The do_fault_around was a few instruction shorter and as far as I can tell saved 12 bytes on the stack, i.e.: $ grep -e rsp -e pop -e push do_fault_around.* do_fault_around.before.s:push %rbp do_fault_around.before.s:mov %rsp,%rbp do_fault_around.before.s:push %r13 do_fault_around.before.s:push %r12 do_fault_around.before.s:push %rbx do_fault_around.before.s:sub $0x38,%rsp do_fault_around.before.s:add $0x38,%rsp do_fault_around.before.s:pop %rbx do_fault_around.before.s:pop %r12 do_fault_around.before.s:pop %r13 do_fault_around.before.s:pop %rbp do_fault_around.after.s:push %rbp do_fault_around.after.s:mov %rsp,%rbp do_fault_around.after.s:push %r12 do_fault_around.after.s:push %rbx do_fault_around.after.s:sub $0x30,%rsp do_fault_around.after.s:add $0x30,%rsp do_fault_around.after.s:pop %rbx do_fault_around.after.s:pop %r12 do_fault_around.after.s:pop %rbp or here side-by-side: Before After push %rbp push %rbp mov %rsp,%rbp mov %rsp,%rbp push %r13 push %r12 push %r12 push %rbx push %rbx sub $0x38,%rsp sub $0x30,%rsp add $0x38,%rsp add $0x30,%rsp pop %rbx pop %rbx pop %r12 pop %r12 pop %r13 pop %rbp pop %rbp There are also fewer branches: $ grep ^j do_fault_around.* do_fault_around.before.s:jae ffffffff812079b7 do_fault_around.before.s:jmp ffffffff812079c5 do_fault_around.before.s:jmp ffffffff81207a14 do_fault_around.before.s:ja ffffffff812079f9 do_fault_around.before.s:jb ffffffff81207a10 do_fault_around.before.s:jmp ffffffff81207a63 do_fault_around.before.s:jne ffffffff812079df do_fault_around.after.s:jmp ffffffff812079fd do_fault_around.after.s:ja ffffffff812079e2 do_fault_around.after.s:jb ffffffff812079f9 do_fault_around.after.s:jmp ffffffff81207a4c do_fault_around.after.s:jne ffffffff812079c8 And here's with allyesconfig on a different machine: $ uname -a; gcc --version; ls -l vmlinux.* Linux erwin 3.14.7-mn #54 SMP Sun Jun 15 11:25:08 CEST 2014 x86_64 AMD Phenom(tm) II X3 710 Processor AuthenticAMD GNU/Linux gcc (GCC) 4.8.3 Copyright (C) 2013 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. -rwx------ 1 mpn eng 437027411 Jun 20 16:04 vmlinux.before -rwx------ 1 mpn eng 437026881 Jun 20 15:30 vmlinux.after 530 bytes reduction. Signed-off-by: Michal Nazarewicz <[email protected]> Signed-off-by: Hagen Paul Pfeifer <[email protected]> Acked-by: Steven Rostedt <[email protected]> Cc: Hagen Paul Pfeifer <[email protected]> Cc: David Rientjes <[email protected]> Cc: "Rustad, Mark D" <[email protected]> Signed-off-by: Andrew Morton <[email protected]> Signed-off-by: Linus Torvalds <[email protected]>

Otherwise rcu_irq_{enter,exit}() do not happen and we get dumps like: ==================== [ 188.275021] =============================== [ 188.309351] [ INFO: suspicious RCU usage. ] [ 188.343737] 3.18.0-rc3-00068-g20f3963-dirty torvalds#54 Not tainted [ 188.394786] ------------------------------- [ 188.429170] include/linux/rcupdate.h:883 rcu_read_lock() used illegally while idle! [ 188.505235] other info that might help us debug this: [ 188.554230] RCU used illegally from idle CPU! rcu_scheduler_active = 1, debug_locks = 0 [ 188.637587] RCU used illegally from extended quiescent state! [ 188.690684] 3 locks held by swapper/7/0: [ 188.721932] #0: (&x->wait#11){......}, at: [<0000000000495de8>] complete+0x8/0x60 [ 188.797994] #1: (&p->pi_lock){-.-.-.}, at: [<000000000048510c>] try_to_wake_up+0xc/0x400 [ 188.881343] #2: (rcu_read_lock){......}, at: [<000000000048a910>] select_task_rq_fair+0x90/0xb40 [ 188.973043]stack backtrace: [ 188.993879] CPU: 7 PID: 0 Comm: swapper/7 Not tainted 3.18.0-rc3-00068-g20f3963-dirty torvalds#54 [ 189.076187] Call Trace: [ 189.089719] [0000000000499360] lockdep_rcu_suspicious+0xe0/0x100 [ 189.147035] [000000000048a99c] select_task_rq_fair+0x11c/0xb40 [ 189.202253] [00000000004852d8] try_to_wake_up+0x1d8/0x400 [ 189.252258] [000000000048554c] default_wake_function+0xc/0x20 [ 189.306435] [0000000000495554] __wake_up_common+0x34/0x80 [ 189.356448] [00000000004955b4] __wake_up_locked+0x14/0x40 [ 189.406456] [0000000000495e08] complete+0x28/0x60 [ 189.448142] [0000000000636e28] blk_end_sync_rq+0x8/0x20 [ 189.496057] [0000000000639898] __blk_mq_end_request+0x18/0x60 [ 189.550249] [00000000006ee014] scsi_end_request+0x94/0x180 [ 189.601286] [00000000006ee334] scsi_io_completion+0x1d4/0x600 [ 189.655463] [00000000006e51c4] scsi_finish_command+0xc4/0xe0 [ 189.708598] [00000000006ed958] scsi_softirq_done+0x118/0x140 [ 189.761735] [00000000006398ec] __blk_mq_complete_request_remote+0xc/0x20 [ 189.827383] [00000000004c75d0] generic_smp_call_function_single_interrupt+0x150/0x1c0 [ 189.906581] [000000000043e514] smp_call_function_single_client+0x14/0x40 ==================== Based almost entirely upon a patch by Paul E. McKenney. Reported-by: Meelis Roos <[email protected]> Tested-by: Meelis Roos <[email protected]> Signed-off-by: David S. Miller <[email protected]>

GIT ca1fa22e67bcfc84f49e44ad6f728e3e3d487dce commit 48eb5b9c3dd2768b6a4de9c1eab606820fd84192 Author: Daniel Borkmann <[email protected]> Date: Tue Nov 11 10:22:05 2014 -0800 ixgbe: phy: fix uninitialized status in ixgbe_setup_phy_link_tnx Status variable is never initialized, can carry an arbitrary value on the stack and thus may let the function fail. Fixes: e90dd2645664 ("ixgbe: Make return values more direct") Signed-off-by: Daniel Borkmann <[email protected]> Acked-by: Emil Tantilov <[email protected]> Signed-off-by: Jeff Kirsher <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit 4eed83a252257ad8ad41ba1c769341960ed4cdc5 Author: James Cameron <[email protected]> Date: Tue Nov 11 16:21:28 2014 +1100 mwifiex: simplify ad hoc join capability info While preparing an ad-hoc start command, the capability info bitmap is needlessly set from the command, and then the ESS bit cleared. Change to set the bitmap directly without reference to the command. Signed-off-by: James Cameron <[email protected]> Acked-by: Amitkumar Karwar <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit d351f5fea44a7527819598070e11b5c9dc53c017 Author: Luciano Coelho <[email protected]> Date: Mon Nov 10 09:25:57 2014 +0200 wlcore: make wlcore_cmd_send_failsafe() static The wlcore_cmd_send_failsafe() function is only called in the cmd.c file, where it is definde. Make it static. Additionally, move the EXPORT_SYMBOL macro for wl1271_cmd_send() to the right place. Signed-off-by: Luciano Coelho <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit f5b8f4790bb5dfd541f9d61589357ea6042cc668 Author: Luciano Coelho <[email protected]> Date: Mon Nov 10 09:25:17 2014 +0200 wlcore: check minimum buffer size in some cmd_send functions Check for the minimum required buffer length in wlcore_cmd_send() and wlcore_cmd_configure_failsafe. This ensures that we will never try to use a buffer that is smaller than the required header. Reported-by: Dan Carpenter <[email protected]> Signed-off-by: Luciano Coelho <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 0d4b5c7c0892cb377cc71c388433425f598b902b Author: Amitkumar Karwar <[email protected]> Date: Fri Nov 7 02:14:52 2014 -0800 mwifiex: fix version display problem on big endian platforms It's been observed that wrong firmware version (ex. 66.14.96.p9 instead of 14.66.9.p96) is displayed on big endian platforms. The problem is fixed here. Reported-by: Daniel Mosquera <[email protected]> Tested-by: Daniel Mosquera <[email protected]> Signed-off-by: Amitkumar Karwar <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 3f2aa13f6d16a53a4cf5de369c685c6f75fe4d58 Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:38 2014 +0100 ath9k_htc: add spectral scan debug interface Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 83fb287ecd8ae60ed79c647a5df8beacdf4f4807 Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:37 2014 +0100 ath9k_htc: process rx spectral packets use code provided by Ashish Patro <[email protected]> Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 911544f6a8d66c27ff807f5d71e3f0f5a904c100 Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:36 2014 +0100 ath9k_htc: trigger spectral scan on set_channel Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 88a2e3fb7b7e9dd3c51ceef81e0415478c27910c Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:35 2014 +0100 ath9k_htc: set initial spec_config values use values provided by Ashish Patro <[email protected]> Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit fe30e8bb5be52e2e6b913db77dda6c22b643dbd7 Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:34 2014 +0100 ath9k_htc: add struct ath_spec_scan_priv to ath9k_htc_priv Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 94cd95c217a5ed4c45e2a8c97043c3efcd66dac9 Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:33 2014 +0100 ath9k_htc: add ath_ps_ops bindings Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 525d09456b9fc2f769647c744c75629d9926fb9e Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:32 2014 +0100 ath9k_htc: fix rs_datalen conversation For some reason it didn't coused obvious problems. Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 46140ddf169703ef0538bf00098233b24b2269e8 Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:31 2014 +0100 ath9k: For AR9271 chipsets, set count = 0 for endless samples. not sure why. Initially provided by Ashish Patro <[email protected]> Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 67dc74f15f147b9f88702de2952d2951e3e000ec Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:30 2014 +0100 ath9k: move spectral.* to common-spectral.* and rename exports from ath9k_spectral_* to ath9k_cmn_spectral_* Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit f00a422cc81ef665f5098c0bc43cb0c616e55a9b Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:29 2014 +0100 ath9k: move ath9k_spectral_scan_ from main.c to spectral.c Now we should be ready to make this code common. Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 963916dfe2907d91eb8a250d12d2b5ae5a1bb343 Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:28 2014 +0100 ath9k: make ath9k_spectral_scan_ do not depend on ath_softc last preparation before moving ath9k_spectral_scan_ to spectral.c Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit ef948da55f20edbb68dac427b7e067c805c852f5 Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:27 2014 +0100 ath9k: use ath_common instead of ieee80211_hw in ath9k_spectral_scan_ we don't have here any ieee80211_hw dependencies any way. Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 934bdc73dd3029c1b91e1a3538268b4afccd58cf Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:26 2014 +0100 ath9k: use ath_ps_ops in ath9k_spectral_scan_ Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 99d2217b731e664aa31001839f12944b1e114a08 Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:25 2014 +0100 ath9k: add ath_ps_ops bindings Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 0198c2e2987c5cd4980f15126d7c68759f4def95 Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:24 2014 +0100 ath: add struct ath_ps_ops we will need it to make common code Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 1111d426ef6a62903a8427a80c2a20cdf0380349 Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:23 2014 +0100 ath9k: remove all struct ath_softc dependencies from spectral code Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit dd7657be756551b23b3431d81e66a8d95a72c923 Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:22 2014 +0100 ath9k: add ath_hw to ath_spec_scan_priv spectral code mostly depends on ath_hw, not on ath_softc Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit c10b75af4344fe0e678d167cb401a94f565e978c Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:21 2014 +0100 ath9k: use struct dentry by ath9k_spectral_init_debug this will alow us to make ath_softc independent code. Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 21af25d00b8bdf03a899b316d41d31ac3eafaf78 Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:20 2014 +0100 ath9k: move spec_config to ath_spec_scan_priv Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 8391f60194bd0d9ab489105381df6455afe1f39a Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:19 2014 +0100 ath9k: move spectral_mode to ath_spec_scan_priv Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 911ea79f435302fabefa305d3649efd4e205672b Author: Oleksij Rempel <[email protected]> Date: Thu Nov 6 08:53:18 2014 +0100 ath9k: add struct ath_spec_scan_priv and move rfs_chan_spec_scan to this struct. We will need it for common spectral scan code. Signed-off-by: Oleksij Rempel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit d7d8b83473e6932cfe9f89e6d839f27abf35b319 Author: Avinash Patil <[email protected]> Date: Wed Nov 5 17:04:31 2014 +0530 mwifiex: fix warning while starting BSS We see this warning while starting mwifiex AP: Unsupported RX-STBC, default to 2x2 This was happening because of wrong offset while copying HT capabilities from BSS configuration of start_ap handler. Signed-off-by: Amitkumar Karwar <[email protected]> Signed-off-by: Avinash Patil <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit bfd713bc1a5d7f01e3d7febe0849b21ae1355c7c Author: Avinash Patil <[email protected]> Date: Wed Nov 5 17:04:30 2014 +0530 mwifiex: do not setup AMPDU/AMSDU with broadcast receiver It is observed that device sometimes sends BA setup requests for broadcast mac address. This patch adds a check to avoid checking availability of AMPDU/AMSDU streams for broadcast mac address. Signed-off-by: Amitkumar Karwar <[email protected]> Signed-off-by: Avinash Patil <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit cf6a64fd603ae0f7391f7589b0f3568d4e79605c Author: Amitkumar Karwar <[email protected]> Date: Wed Nov 5 17:04:29 2014 +0530 mwifiex: fix out of memory issue observed for USB chipsets On some platforms, system goes out of memory during heavy Rx traffic with our USB chipsets. In case of SDIO/PCIe, after receiving 50 packets in Rx queue we stop processing interrupts till packets pending fall below low threshold i.e 20. We don't have similar logic for USB, so if host platform is slow, we would hit a case where firmware keeps on pushing packets at high speed than driver/kernel can process. We will stop submitting URBs for Rx data when pending packet count reaches high threshold and restart them when enough packets are consumed to solve the problem. BugLink: https://bugzilla.kernel.org/show_bug.cgi?id=85071 Reported-by: Marek Belisko <[email protected]> Tested-by: Marek Belisko <[email protected]> Signed-off-by: Avinash Patil <[email protected]> Signed-off-by: Cathy Luo <[email protected]> Signed-off-by: Amitkumar Karwar <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 041bfab5bbb6ec721c743f487e3e22b87f666996 Author: Avinash Patil <[email protected]> Date: Wed Nov 5 17:04:28 2014 +0530 mwifiex: remove data_complete handler This patch removes redundant data complete handler. Signed-off-by: Avinash Patil <[email protected]> Signed-off-by: Cathy Luo <[email protected]> Signed-off-by: Amitkumar Karwar <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit ec4a16b4d287d4d0f7465ae7e61ce4e9021d715c Author: Avinash Patil <[email protected]> Date: Wed Nov 5 17:04:27 2014 +0530 mwifiex: rx workqueue support for USB interface This patch adds RX workqueue support for USB interfaces. Currently rx_pending is applicable for cmd/events and Rx data in USB interface. Let's use it only for Rx data. Signed-off-by: Avinash Patil <[email protected]> Signed-off-by: Cathy Luo <[email protected]> Signed-off-by: Amitkumar Karwar <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit d385c5c2860075e1f3e03074f043dd8a828b2862 Author: Felix Fietkau <[email protected]> Date: Tue Nov 4 16:56:57 2014 +0100 ath9k: add support for reporting tx power to mac80211 Track it per channel context instead of in the softc Signed-off-by: Felix Fietkau <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 71783576b5345d63df048c0f18974037eea6e4f9 Author: Hauke Mehrtens <[email protected]> Date: Sat Nov 1 16:54:56 2014 +0100 bcma: get IRQ numbers from dt It is not possible to auto detect the irq numbers used by the cores on an arm SoC. If bcma was registered with device tree it will search for some device tree nodes with the irq number and add it to the core configuration. Signed-off-by: Hauke Mehrtens <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 85eb92e81801d64686eb78928d500a4c83ee9623 Author: Hauke Mehrtens <[email protected]> Date: Sat Nov 1 16:54:55 2014 +0100 bcma: make it possible to specify a IRQ num in bcma_core_irq() This moves bcma_core_irq() to main.c and add a extra parameter with a number so that we can return different irq number for devices with more than one. Signed-off-by: Hauke Mehrtens <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 09626e9d153326ca82568e4e27f2daa53713992e Author: WANG Cong <[email protected]> Date: Tue Nov 11 13:29:42 2014 -0800 net: kill netif_copy_real_num_queues() vlan was the only user of netif_copy_real_num_queues(), but it no longer calls it after commit 4af429d29b341bb1735f04c2fb960178 ("vlan: lockless transmit path"). So we can just remove it. Cc: Eric Dumazet <[email protected]> Cc: David S. Miller <[email protected]> Signed-off-by: Cong Wang <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit 67732cd34382066ae5df313b6dad65ab14b9735f Author: Ulf Hansson <[email protected]> Date: Tue Nov 11 11:07:08 2014 +0100 PM / Domains: Fix initial default state of the need_restore flag The initial state of the device's need_restore flag should'nt depend on the current state of the PM domain. For example it should be perfectly valid to attach an inactive device to a powered PM domain. The pm_genpd_dev_need_restore() API allow us to update the need_restore flag to somewhat cope with such scenarios. Typically that should have been done from drivers/buses ->probe() since it's those that put the requirements on the value of the need_restore flag. Until recently, the Exynos SOCs were the only user of the pm_genpd_dev_need_restore() API, though invoking it from a centralized location while adding devices to their PM domains. Due to that Exynos now have swithed to the generic OF-based PM domain look-up, it's no longer possible to invoke the API from a centralized location. The reason is because devices are now added to their PM domains during the probe sequence. Commit "ARM: exynos: Move to generic PM domain DT bindings" did the switch for Exynos to the generic OF-based PM domain look-up, but it also removed the call to pm_genpd_dev_need_restore(). This caused a regression for some of the Exynos drivers. To handle things more properly in the generic PM domain, let's change the default initial value of the need_restore flag to reflect that the state is unknown. As soon as some of the runtime PM callbacks gets invoked, update the initial value accordingly. Moreover, since the generic PM domain is verifying that all devices are both runtime PM enabled and suspended, using pm_runtime_suspended() while pm_genpd_poweroff() is invoked from the scheduled work, we can be sure of that the PM domain won't be powering off while having active devices. Do note that, the generic PM domain can still only know about active devices which has been activated through invoking its runtime PM resume callback. In other words, buses/drivers using pm_runtime_set_active() during ->probe() will still suffer from a race condition, potentially probing a device without having its PM domain being powered. That issue will have to be solved using a different approach. This a log from the boot regression for Exynos5, which is being fixed in this patch. ------------[ cut here ]------------ WARNING: CPU: 0 PID: 308 at ../drivers/clk/clk.c:851 clk_disable+0x24/0x30() Modules linked in: CPU: 0 PID: 308 Comm: kworker/0:1 Not tainted 3.18.0-rc3-00569-gbd9449f-dirty #10 Workqueue: pm pm_runtime_work [<c0013c64>] (unwind_backtrace) from [<c0010dec>] (show_stack+0x10/0x14) [<c0010dec>] (show_stack) from [<c03ee4cc>] (dump_stack+0x70/0xbc) [<c03ee4cc>] (dump_stack) from [<c0020d34>] (warn_slowpath_common+0x64/0x88) [<c0020d34>] (warn_slowpath_common) from [<c0020d74>] (warn_slowpath_null+0x1c/0x24) [<c0020d74>] (warn_slowpath_null) from [<c03107b0>] (clk_disable+0x24/0x30) [<c03107b0>] (clk_disable) from [<c02cc834>] (gsc_runtime_suspend+0x128/0x160) [<c02cc834>] (gsc_runtime_suspend) from [<c0249024>] (pm_generic_runtime_suspend+0x2c/0x38) [<c0249024>] (pm_generic_runtime_suspend) from [<c024f44c>] (pm_genpd_default_save_state+0x2c/0x8c) [<c024f44c>] (pm_genpd_default_save_state) from [<c024ff2c>] (pm_genpd_poweroff+0x224/0x3ec) [<c024ff2c>] (pm_genpd_poweroff) from [<c02501b4>] (pm_genpd_runtime_suspend+0x9c/0xcc) [<c02501b4>] (pm_genpd_runtime_suspend) from [<c024a4f8>] (__rpm_callback+0x2c/0x60) [<c024a4f8>] (__rpm_callback) from [<c024a54c>] (rpm_callback+0x20/0x74) [<c024a54c>] (rpm_callback) from [<c024a930>] (rpm_suspend+0xd4/0x43c) [<c024a930>] (rpm_suspend) from [<c024bbcc>] (pm_runtime_work+0x80/0x90) [<c024bbcc>] (pm_runtime_work) from [<c0032a9c>] (process_one_work+0x12c/0x314) [<c0032a9c>] (process_one_work) from [<c0032cf4>] (worker_thread+0x3c/0x4b0) [<c0032cf4>] (worker_thread) from [<c003747c>] (kthread+0xcc/0xe8) [<c003747c>] (kthread) from [<c000e738>] (ret_from_fork+0x14/0x3c) ---[ end trace 40cd58bcd6988f12 ]--- Fixes: a4a8c2c4962bb655 (ARM: exynos: Move to generic PM domain DT bindings) Reported-and-tested0by: Sylwester Nawrocki <[email protected]> Reviewed-by: Sylwester Nawrocki <[email protected]> Reviewed-by: Kevin Hilman <[email protected]> Signed-off-by: Ulf Hansson <[email protected]> Signed-off-by: Rafael J. Wysocki <[email protected]> commit 4e6ce4dc7ce71d0886908d55129d5d6482a27ff9 Author: Miaoqing Pan <[email protected]> Date: Thu Nov 6 10:52:23 2014 +0530 ath9k: Fix RTC_DERIVED_CLK usage Based on the reference clock, which could be 25MHz or 40MHz, AR_RTC_DERIVED_CLK is programmed differently for AR9340 and AR9550. But, when a chip reset is done, processing the initvals sets the register back to the default value. Fix this by moving the code in ath9k_hw_init_pll() to ar9003_hw_override_ini(). Also, do this override for AR9531. Cc: [email protected] Signed-off-by: Miaoqing Pan <[email protected]> Signed-off-by: Sujith Manoharan <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit 8bca81d9875c7768c40a19fb439eebaf6cec898d Author: Sudip Mukherjee <[email protected]> Date: Tue Nov 11 14:10:47 2014 +0530 usbnet: smsc95xx: dereferencing NULL pointer we were dereferencing dev to initialize pdata. but just after that we have a BUG_ON(!dev). so we were basically dereferencing the pointer first and then tesing it for NULL. Signed-off-by: Sudip Mukherjee <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit d65c4e4e0aeb699e984bd4b382efffab418aa359 Author: Joe Perches <[email protected]> Date: Tue Nov 11 13:13:41 2014 -0800 irda: Simplify IRDA logging macros These are the same as net_<level>_ratelimited, so use the more common style in the macro definition. Signed-off-by: Joe Perches <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit 79ce0477ffe82e7e49e55179cd176a1c33382744 Author: Brian Hill <[email protected]> Date: Tue Nov 11 13:39:39 2014 -0700 net: phy: Correctly handle MII ioctl which changes autonegotiation. When advertised capabilities are changed with mii-tool, such as: mii-tool -A 10baseT the existing handler has two errors. - An actual PHY register value is provided by mii-tool, and this must be mapped to internal state with mii_adv_to_ethtool_adv_t(). - The PHY state machine needs to be told that autonegotiation has again been performed. If not, the MAC will not be notified of the new link speed and duplex, resulting in a possible config mismatch. Signed-off-by: Brian Hill <[email protected]> Acked-by: Florian Fainelli <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit 0cd75b19899fd86b51a6480fb8c00dcd85a54591 Author: Arend van Spriel <[email protected]> Date: Tue Nov 11 13:58:44 2014 +0100 brcmfmac: fix conversion of channel width 20MHZ_NOHT The function chandef_to_chanspec() failed when converting a chandef with bandwidth set to NL80211_CHAN_WIDTH_20_NOHT. This was reported by user running the device in AP mode. ------------[ cut here ]------------ WARNING: CPU: 0 PID: 304 at drivers/net/wireless/brcm80211/brcmfmac/wl_cfg80211.c:381 chandef_to_chanspec.isra.11+0x158/0x184() Modules linked in: CPU: 0 PID: 304 Comm: hostapd Not tainted 3.16.0-rc7-abb+g64aa90f #8 [<c0014bb4>] (unwind_backtrace) from [<c0012314>] (show_stack+0x10/0x14) [<c0012314>] (show_stack) from [<c001d3f8>] (warn_slowpath_common+0x6c/0x8c) [<c001d3f8>] (warn_slowpath_common) from [<c001d4b4>] (warn_slowpath_null+0x1c/0x24) [<c001d4b4>] (warn_slowpath_null) from [<c03449a4>] (chandef_to_chanspec.isra.11+0x158/0x184) [<c03449a4>] (chandef_to_chanspec.isra.11) from [<c0348e00>] (brcmf_cfg80211_start_ap+0x1e4/0x614) [<c0348e00>] (brcmf_cfg80211_start_ap) from [<c04d1468>] (nl80211_start_ap+0x288/0x414) [<c04d1468>] (nl80211_start_ap) from [<c043d144>] (genl_rcv_msg+0x21c/0x38c) [<c043d144>] (genl_rcv_msg) from [<c043c740>] (netlink_rcv_skb+0xac/0xc0) [<c043c740>] (netlink_rcv_skb) from [<c043cf14>] (genl_rcv+0x20/0x34) [<c043cf14>] (genl_rcv) from [<c043c0a0>] (netlink_unicast+0x150/0x20c) [<c043c0a0>] (netlink_unicast) from [<c043c4b8>] (netlink_sendmsg+0x2b8/0x398) [<c043c4b8>] (netlink_sendmsg) from [<c04066a4>] (sock_sendmsg+0x84/0xa8) [<c04066a4>] (sock_sendmsg) from [<c0407c5c>] (___sys_sendmsg.part.29+0x268/0x278) [<c0407c5c>] (___sys_sendmsg.part.29) from [<c0408bdc>] (__sys_sendmsg+0x4c/0x7c) [<c0408bdc>] (__sys_sendmsg) from [<c000ec60>] (ret_fast_syscall+0x0/0x44) ---[ end trace 965ee2158c9905a2 ]--- Cc: [email protected] # v3.17 Reported-by: Pontus Fuchs <[email protected]> Reviewed-by: Hante Meuleman <[email protected]> Reviewed-by: Daniel (Deognyoun) Kim <[email protected]> Reviewed-by: Franky (Zhenhui) Lin <[email protected]> Reviewed-by: Pieter-Paul Giesberts <[email protected]> Signed-off-by: Arend van Spriel <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit cfd9167af14eb4ec21517a32911d460083ee3d59 Author: Stanislaw Gruszka <[email protected]> Date: Tue Nov 11 14:28:47 2014 +0100 rt2x00: do not align payload on modern H/W RT2800 and newer hardware require padding between header and payload if header length is not multiple of 4. For historical reasons we also align payload to to 4 bytes boundary, but such alignment is not needed on modern H/W. Patch fixes skb_under_panic problems reported from time to time: https://bugzilla.kernel.org/show_bug.cgi?id=84911 https://bugzilla.kernel.org/show_bug.cgi?id=72471 http://marc.info/?l=linux-wireless&m=139108549530402&w=2 https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1087591 Panic happened because we eat 4 bytes of skb headroom on each (re)transmission when sending frame without the payload and the header length not being multiple of 4 (i.e. QoS header has 26 bytes). On such case because paylad_aling=2 is bigger than header_align=0 we increase header_align by 4 bytes. To prevent that we could change the check to: if (payload_length && payload_align > header_align) header_align += 4; but not aligning payload at all is more effective and alignment is not really needed by H/W (that has been tested on OpenWrt project for few years now). Reported-and-tested-by: Antti S. Lankila <[email protected]> Debugged-by: Antti S. Lankila <[email protected]> Reported-by: Henrik Asp <[email protected]> Originally-From: Helmut Schaa <[email protected]> Cc: [email protected] Signed-off-by: Stanislaw Gruszka <[email protected]> Signed-off-by: John W. Linville <[email protected]> commit f47436734dc89ece62654d4db8d08163a89dd7ca Author: Joe Perches <[email protected]> Date: Fri Oct 31 10:50:46 2014 -0700 tile: Use the more common pr_warn instead of pr_warning And other message logging neatening. Other miscellanea: o coalesce formats o realign arguments o standardize a couple of macros o use __func__ instead of embedding the function name Signed-off-by: Joe Perches <[email protected]> Signed-off-by: Chris Metcalf <[email protected]> commit ebd25caf7d511312d1a9724ab5752e9e661dfe60 Author: Chen Gang <[email protected]> Date: Sun Nov 9 18:32:03 2014 +0800 arch: tile: gxio: Export symbols for module using in 'mpipe.c' 'gxio_mpipe_adjust_timestamp', 'gxio_mpipe_link_instance', 'gxio_mpipe_get_timestamp', and 'gxio_mpipe_set_timestamp' may be use by other tile modules, so export them. The related error (with allmodconfig under tile): MODPOST 4002 modules ERROR: "gxio_mpipe_link_instance" [drivers/net/ethernet/tile/tile_net.ko] undefined! ERROR: "gxio_mpipe_get_timestamp" [drivers/net/ethernet/tile/tile_net.ko] undefined! ERROR: "gxio_mpipe_set_timestamp" [drivers/net/ethernet/tile/tile_net.ko] undefined! ERROR: "gxio_mpipe_adjust_timestamp" [drivers/net/ethernet/tile/tile_net.ko] undefined! Signed-off-by: Chen Gang <[email protected]> Signed-off-by: Chris Metcalf <[email protected]> commit 5337b5b75cd9bd3624a6820e3c2a084d2480061c Author: Eric Dumazet <[email protected]> Date: Mon Nov 10 17:54:25 2014 -0800 ipv6: fix IPV6_PKTINFO with v4 mapped Use IS_ENABLED(CONFIG_IPV6), to enable this code if IPv6 is a module. Signed-off-by: Eric Dumazet <[email protected]> Fixes: c8e6ad0829a7 ("ipv6: honor IPV6_PKTINFO with v4 mapped addresses on sendmsg") Acked-by: Hannes Frederic Sowa <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit d7480fd3b1738a8eae6a76098b17af318cf9b9cc Author: WANG Cong <[email protected]> Date: Mon Nov 10 15:59:36 2014 -0800 neigh: remove dynamic neigh table registration support Currently there are only three neigh tables in the whole kernel: arp table, ndisc table and decnet neigh table. What's more, we don't support registering multiple tables per family. Therefore we can just make these tables statically built-in. Cc: David S. Miller <[email protected]> Signed-off-by: Cong Wang <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit 4184b2a79a7612a9272ce20d639934584a1f3786 Author: Daniel Borkmann <[email protected]> Date: Mon Nov 10 18:00:09 2014 +0100 net: sctp: fix memory leak in auth key management A very minimal and simple user space application allocating an SCTP socket, setting SCTP_AUTH_KEY setsockopt(2) on it and then closing the socket again will leak the memory containing the authentication key from user space: unreferenced object 0xffff8800837047c0 (size 16): comm "a.out", pid 2789, jiffies 4296954322 (age 192.258s) hex dump (first 16 bytes): 01 00 00 00 04 00 00 00 00 00 00 00 00 00 00 00 ................ backtrace: [<ffffffff816d7e8e>] kmemleak_alloc+0x4e/0xb0 [<ffffffff811c88d8>] __kmalloc+0xe8/0x270 [<ffffffffa0870c23>] sctp_auth_create_key+0x23/0x50 [sctp] [<ffffffffa08718b1>] sctp_auth_set_key+0xa1/0x140 [sctp] [<ffffffffa086b383>] sctp_setsockopt+0xd03/0x1180 [sctp] [<ffffffff815bfd94>] sock_common_setsockopt+0x14/0x20 [<ffffffff815beb61>] SyS_setsockopt+0x71/0xd0 [<ffffffff816e58a9>] system_call_fastpath+0x12/0x17 [<ffffffffffffffff>] 0xffffffffffffffff This is bad because of two things, we can bring down a machine from user space when auth_enable=1, but also we would leave security sensitive keying material in memory without clearing it after use. The issue is that sctp_auth_create_key() already sets the refcount to 1, but after allocation sctp_auth_set_key() does an additional refcount on it, and thus leaving it around when we free the socket. Fixes: 65b07e5d0d0 ("[SCTP]: API updates to suport SCTP-AUTH extensions.") Signed-off-by: Daniel Borkmann <[email protected]> Cc: Vlad Yasevich <[email protected]> Acked-by: Neil Horman <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit e40607cbe270a9e8360907cb1e62ddf0736e4864 Author: Daniel Borkmann <[email protected]> Date: Mon Nov 10 17:54:26 2014 +0100 net: sctp: fix NULL pointer dereference in af->from_addr_param on malformed packet An SCTP server doing ASCONF will panic on malformed INIT ping-of-death in the form of: ------------ INIT[PARAM: SET_PRIMARY_IP] ------------> While the INIT chunk parameter verification dissects through many things in order to detect malformed input, it misses to actually check parameters inside of parameters. E.g. RFC5061, section 4.2.4 proposes a 'set primary IP address' parameter in ASCONF, which has as a subparameter an address parameter. So an attacker may send a parameter type other than SCTP_PARAM_IPV4_ADDRESS or SCTP_PARAM_IPV6_ADDRESS, param_type2af() will subsequently return 0 and thus sctp_get_af_specific() returns NULL, too, which we then happily dereference unconditionally through af->from_addr_param(). The trace for the log: BUG: unable to handle kernel NULL pointer dereference at 0000000000000078 IP: [<ffffffffa01e9c62>] sctp_process_init+0x492/0x990 [sctp] PGD 0 Oops: 0000 [#1] SMP [...] Pid: 0, comm: swapper Not tainted 2.6.32-504.el6.x86_64 #1 Bochs Bochs RIP: 0010:[<ffffffffa01e9c62>] [<ffffffffa01e9c62>] sctp_process_init+0x492/0x990 [sctp] [...] Call Trace: <IRQ> [<ffffffffa01f2add>] ? sctp_bind_addr_copy+0x5d/0xe0 [sctp] [<ffffffffa01e1fcb>] sctp_sf_do_5_1B_init+0x21b/0x340 [sctp] [<ffffffffa01e3751>] sctp_do_sm+0x71/0x1210 [sctp] [<ffffffffa01e5c09>] ? sctp_endpoint_lookup_assoc+0xc9/0xf0 [sctp] [<ffffffffa01e61f6>] sctp_endpoint_bh_rcv+0x116/0x230 [sctp] [<ffffffffa01ee986>] sctp_inq_push+0x56/0x80 [sctp] [<ffffffffa01fcc42>] sctp_rcv+0x982/0xa10 [sctp] [<ffffffffa01d5123>] ? ipt_local_in_hook+0x23/0x28 [iptable_filter] [<ffffffff8148bdc9>] ? nf_iterate+0x69/0xb0 [<ffffffff81496d10>] ? ip_local_deliver_finish+0x0/0x2d0 [<ffffffff8148bf86>] ? nf_hook_slow+0x76/0x120 [<ffffffff81496d10>] ? ip_local_deliver_finish+0x0/0x2d0 [...] A minimal way to address this is to check for NULL as we do on all other such occasions where we know sctp_get_af_specific() could possibly return with NULL. Fixes: d6de3097592b ("[SCTP]: Add the handling of "Set Primary IP Address" parameter to INIT") Signed-off-by: Daniel Borkmann <[email protected]> Cc: Vlad Yasevich <[email protected]> Acked-by: Neil Horman <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit 5748eb8f8e989a9da1ac7c96dc73d68cbdedf7df Author: Takashi Iwai <[email protected]> Date: Mon Nov 10 11:50:21 2014 +0100 net: ppp: Don't call bpf_prog_create() in ppp_lock In ppp_ioctl(), bpf_prog_create() is called inside ppp_lock, which eventually calls vmalloc() and hits BUG_ON() in vmalloc.c. This patch works around the problem by moving the allocation outside the lock. The bug was revealed by the recent change in net/core/filter.c, as it allocates via vmalloc() instead of kmalloc() now. Reported-and-tested-by: Stefan Seyfried <[email protected]> Signed-off-by: Takashi Iwai <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit b2e2f0c779fefede3a871781c8827bd8e76c7c0f Author: Andy Shevchenko <[email protected]> Date: Mon Nov 10 12:38:59 2014 +0200 stmmac: split to core library and probe drivers Instead of registering the platform and PCI drivers in one module let's move necessary bits to where it belongs. During this procedure we convert the module registration part to use module_*_driver() macros which makes code simplier. >From now on the driver consists three parts: core library, PCI, and platform drivers. Signed-off-by: Andy Shevchenko <[email protected]> Acked-by: Giuseppe Cavallaro <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit dc680b989d519952e0e0d37204fec850925a0225 Author: Linus Walleij <[email protected]> Date: Mon Nov 10 18:52:31 2014 +0100 ARM: fix multiplatform allmodcompile Commit 68f3b875f7848f5304472184a4634148c5330cbd "ARM: integrator: make the Integrator multiplatform" broke allmodconfig like this: >> arch/arm/include/asm/cmpxchg.h:114:2: error: #error "SMP is not supported on this platform" (etc) This is due to the fact that as we turned on multiplatform for the Integrator, this enabled a lot of non-applicable CPU's to be selected for its multiplatform images, due to a lot of "depends on ARCH_INTEGRATOR" restrictions in arch/arm/mm/Kconfig for the different ARM CPU types. Fix this by restricting the CPU selections to respective multiplatform config, which now becomes a subset of the possible Integrator configurations, or alternatively the non-multiplatform config plus ARCH_INTEGRATOR, i.e.: if (!ARCH_MULTIPLATFORM || ARCH_MULTI_Vx) && (ARCH_INTEGRATOR || ARCH_FOO ...) Since the Integrator has been converted to multiplatform, this will often take the short form: if (ARCH_MULTI_Vx && ARCH_INTEGRATOR) If no other non-multiplatform platforms are elegible. Reported-by: Build bot for Mark Brown <[email protected]> Reported-by: Kbuild test robot <[email protected]> Suggested-by: Russell King <[email protected]> Signed-off-by: Linus Walleij <[email protected]> Signed-off-by: Arnd Bergmann <[email protected]> commit ba7a46f16dd29f93303daeb1fee8af316c5a07f4 Author: Joe Perches <[email protected]> Date: Tue Nov 11 10:59:17 2014 -0800 net: Convert LIMIT_NETDEBUG to net_dbg_ratelimited Use the more common dynamic_debug capable net_dbg_ratelimited and remove the LIMIT_NETDEBUG macro. All messages are still ratelimited. Some KERN_<LEVEL> uses are changed to KERN_DEBUG. This may have some negative impact on messages that were emitted at KERN_INFO that are not not enabled at all unless DEBUG is defined or dynamic_debug is enabled. Even so, these messages are now _not_ emitted by default. This also eliminates the use of the net_msg_warn sysctl "/proc/sys/net/core/warnings". For backward compatibility, the sysctl is not removed, but it has no function. The extern declaration of net_msg_warn is removed from sock.h and made static in net/core/sysctl_net_core.c Miscellanea: o Update the sysctl documentation o Remove the embedded uses of pr_fmt o Coalesce format fragments o Realign arguments Signed-off-by: Joe Perches <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit e349d9d5b3f9509e6a053d9d1d9f4c7d9471c8f0 Author: Neelesh Gupta <[email protected]> Date: Wed Nov 5 16:45:14 2014 +0530 hwmon: (ibmpowernv) Use platform 'id_table' to probe the device The current driver probe() function assumes the sensor device to be always present and gets executed every time if the driver is loaded, but the appropriate hardware could not be present. So, move the platform device creation as part of platform init code and use the 'id_table' to check if the device is present or not. Signed-off-by: Neelesh Gupta <[email protected]> Acked-by: Michael Ellerman <[email protected]> Signed-off-by: Guenter Roeck <[email protected]> commit eebb2554a0a4ce8f637c72febc883997a58093ca Author: Guenter Roeck <[email protected]> Date: Sat Sep 27 08:31:12 2014 -0700 hwmon: (iio_hwmon) Add support for humidity sensors The iio subsystem supports humidity sensors, so it makes sense to support it in the iio-hwmon bridge as well. Cc: Jonathan Cameron <[email protected]> Acked-by: Jonathan Cameron <[email protected]> Signed-off-by: Guenter Roeck <[email protected]> commit c2827f34af5ced95daddefd7de6105d7fcf0d4d5 Author: Alan Tull <[email protected]> Date: Wed Oct 15 13:55:10 2014 -0500 hwmon: (ltc2978) Add regulator support Add simple on/off regulator support for ltc2978 and other pmbus parts supported by the ltc2978 driver. Signed-off-by: Alan Tull <[email protected]> Cc: Guenter Roeck <[email protected]> Cc: Mark Brown <[email protected]> Signed-off-by: Guenter Roeck <[email protected]> commit 3ca7aa3feedc0cc838b242e09ff6b306501029d0 Author: Alan Tull <[email protected]> Date: Wed Oct 15 13:55:09 2014 -0500 hwmon: (pmbus) Add regulator support Add support for simple on/off control of each channel. To add regulator support, the pmbus part driver needs to add regulator_desc information and number of regulators to its pmbus_driver_info struct. regulator_desc can be declared using default macro for a regulator (PMBUS_REGULATOR) that is in pmbus.h The regulator_init_data can be initialized from either platform data or the device tree. Signed-off-by: Alan Tull <[email protected]> Reviewed-by: Mark Brown <[email protected]> Signed-off-by: Guenter Roeck <[email protected]> commit e94c450610a10a9b798b39754a6099fea5d5279e Author: Alan Tull <[email protected]> Date: Wed Oct 15 13:55:08 2014 -0500 hwmon: (pmbus) add helpers for byte write and read modify write Add two helper functions: * pmbus_write_byte_data = paged byte write * pmbus_update_byte_data = paged byte read/modify/write Signed-off-by: Alan Tull <[email protected]> Cc: Mark Brown <[email protected]> Signed-off-by: Guenter Roeck <[email protected]> commit 7c6d297620427048742977c2258669f3cc926f1f Author: Alan Tull <[email protected]> Date: Wed Oct 15 13:55:07 2014 -0500 hwmon: (ltc2978) device tree bindings documentation Add device tree bindings documentation for ltc2978. Signed-off-by: Alan Tull <[email protected]> Cc: Mark Rutland <[email protected]> Cc: Mark Brown <[email protected]> [Guenter Roeck: Minor correction of 'compatible' example] Signed-off-by: Guenter Roeck <[email protected]> commit 5b61c4db49e2530ed10631321d4c73f49d560a93 Author: Denis Kirjanov <[email protected]> Date: Mon Nov 10 08:59:43 2014 +0300 PPC: bpf_jit_comp: add SKF_AD_HATYPE instruction Add BPF extension SKF_AD_HATYPE to ppc JIT to check the hw type of the interface Before: [ 57.723666] test_bpf: #20 LD_HATYPE [ 57.723675] BPF filter opcode 0020 (@0) unsupported [ 57.724168] 48 48 PASS After: [ 103.053184] test_bpf: #20 LD_HATYPE 7 6 PASS CC: Alexei Starovoitov<[email protected]> CC: Daniel Borkmann<[email protected]> CC: Philippe Bergheaud<[email protected]> Signed-off-by: Denis Kirjanov <[email protected]> v2: address Alexei's comments Acked-by: Alexei Starovoitov <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit 0bd52941586b3b59ab9b6e89e55b2dc9e2680de9 Author: Aravind Gopalakrishnan <[email protected]> Date: Tue Nov 4 11:49:02 2014 -0600 hwmon: (fam15h_power) Fix NB device ID for F16h M30h F3 device ID is wrongly included in fam15h_power_id_table for F16h M30h. It should be F4 device ID. Fix this. Signed-off-by: Aravind Gopalakrishnan <[email protected]> Signed-off-by: Guenter Roeck <[email protected]> commit 48b9d5b4f408259cd6800c4b17d4fe5025435da2 Author: Kamil Debski <[email protected]> Date: Mon Nov 3 15:42:55 2014 +0100 hwmon: (pwm-fan) Fix suspend/resume behavior The state of a PWM output is not clearly defined after resume. Some PWM drivers do not restore the duty cycle upon resume, thus it is necessary to manually restore the correct value. Signed-off-by: Kamil Debski <[email protected]> Signed-off-by: Guenter Roeck <[email protected]> commit aab18da44f243cf59b4dee335ea50b32f529b5b0 Author: Michael Ellerman <[email protected]> Date: Fri Oct 31 17:45:22 2014 +1100 hwmon: (ibmpowernv) Quieten when probing finds no device Because we build kernels with drivers built in for many platforms, it's normal for the ibmpowernv driver to be loaded on systems that don't have the appropriate hardware. Currently the driver spams the log with: ibmpowernv ibmpowernv.0: Opal node 'sensors' not found ibmpowernv: Platfrom driver probe failed But there is no error, this machine is not a powernv and doesn't have the hardware. So change the sensors message to dev_dbg(), and only print an error about the probe failing if it's not ENODEV. Also fix the spelling of "Platfrom" and print the actual error value. Signed-off-by: Michael Ellerman <[email protected]> Reviewed-by: Jean Delvare <[email protected]> Signed-off-by: Guenter Roeck <[email protected]> commit a2ae6007a442d6bb27d77bf20ec1b06cda9e306e Author: Joe Perches <[email protected]> Date: Sun Nov 9 16:32:46 2014 -0800 dsa: Use netdev_<level> instead of printk Neaten and standardize the logging output. Other miscellanea: o Use pr_notice_once instead of a guard flag. o Convert existing pr_<level> uses too. Signed-off-by: Joe Perches <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit f4a1edd56120249198073aa4a373b77e3700ac8f Author: Or Gerlitz <[email protected]> Date: Sun Nov 9 14:25:39 2014 +0200 net/mlx4_en: Advertize encapsulation offloads features only when VXLAN tunnel is set Currenly we only support Large-Send and TX checksum offloads for encapsulated traffic of type VXLAN. We must make sure to advertize these offloads up to the stack only when VXLAN tunnel is set. Failing to do so, would mislead the the networking stack to assume that the driver can offload the internal TX checksum for GRE packets and other buggy schemes. Reported-by: Florian Westphal <[email protected]> Signed-off-by: Or Gerlitz <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit f8c6455bb04b944edb69e9b074e28efee2c56bdd Author: Shani Michaeli <[email protected]> Date: Sun Nov 9 13:51:53 2014 +0200 net/mlx4_en: Extend checksum offloading by CHECKSUM COMPLETE When processing received traffic, pass CHECKSUM_COMPLETE status to the stack, with calculated checksum for non TCP/UDP packets (such as GRE or ICMP). Although the stack expects checksum which doesn't include the pseudo header, the HW adds it. To address that, we are subtracting the pseudo header checksum from the checksum value provided by the HW. In the IPv6 case, we also compute/add the IP header checksum which is not added by the HW for such packets. Cc: Jerry Chu <[email protected]> Signed-off-by: Shani Michaeli <[email protected]> Signed-off-by: Matan Barak <[email protected]> Signed-off-by: Or Gerlitz <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit dd65beac48a5259945846956d4b27344dfb73bd9 Author: Shani Michaeli <[email protected]> Date: Sun Nov 9 13:51:52 2014 +0200 net/mlx4_en: Extend usage of napi_gro_frags We can call napi_gro_frags for all the received traffic regardless of the checksum status. Specifically, received packets whose status is CHECKSUM_NONE (and soon to be added CHECKSUM_COMPLETE) are eligible for napi_gro_frags as well. Signed-off-by: Or Gerlitz <[email protected]> Signed-off-by: Shani Michaeli <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit 242fe7a1fd7af133d4c0ae2aae1de4e33b1b39af Author: Rafał Miłecki <[email protected]> Date: Wed Sep 3 07:36:51 2014 +0200 MIPS: Fix info about plat_setup in arch_mem_init comment Signed-off-by: Rafał Miłecki <[email protected]> Cc: [email protected] Patchwork: https://patchwork.linux-mips.org/patch/7607/ Signed-off-by: Ralf Baechle <[email protected]> commit 5996d33df117bc3c56c28d6a930679ddcb693626 Author: Rafał Miłecki <[email protected]> Date: Thu Oct 30 12:50:03 2014 +0100 MIPS: BCM47XX: Clean up nvram header 1) Move private defines to the .c file 2) Move SPROM helper to the sprom.c 3) Drop unused code 4) Rename magic to the NVRAM_MAGIC 5) Add const to the char pointer we never modify Signed-off-by: Rafał Miłecki <[email protected]> Acked-by: Hauke Mehrtens <[email protected]> Cc: [email protected] Patchwork: https://patchwork.linux-mips.org/patch/8289/ Signed-off-by: Ralf Baechle <[email protected]> commit 1bb002b25f08e7e295b2c16dfc09a4cefc3cc5db Author: Rafał Miłecki <[email protected]> Date: Wed Oct 29 10:05:06 2014 +0100 MIPS: BCM47XX: Use mtd as an alternative way/API to get NVRAM content NVRAM can be read using magic memory offset, but after all it's just a flash partition. On platforms where NVRAM isn't needed early we can get it using mtd subsystem. Signed-off-by: Rafał Miłecki <[email protected]> Acked-by: Hauke Mehrtens <[email protected]> Cc: [email protected] Patchwork: https://patchwork.linux-mips.org/patch/8266/ Signed-off-by: Ralf Baechle <[email protected]> commit 2a90d4aae5509e9cf1ba848c5d0b3458201160a0 Author: Paolo Bonzini <[email protected]> Date: Fri Nov 7 23:04:00 2014 +0100 blk-mq: use get_cpu/put_cpu instead of preempt_disable/preempt_enable blk-mq is using preempt_disable/enable in order to ensure that the queue runners are placed on the right CPU. This does not work with the RT patches, because __blk_mq_run_hw_queue takes a non-raw spinlock with the preemption-disabled region. If there is contention on the lock, this violates the rules for preemption-disabled regions. While this should be easily fixable within the RT patches just by doing migrate_disable/enable, we can do better and document _why_ this particular region runs with disabled preemption. After the previous patch, it is trivial to switch it to get/put_cpu; the RT patches then can change it to get_cpu_light, which lets virtio-blk run under RT kernels. Cc: Jens Axboe <[email protected]> Cc: Thomas Gleixner <[email protected]> Reported-by: Clark Williams <[email protected]> Tested-by: Clark Williams <[email protected]> Signed-off-by: Paolo Bonzini <[email protected]> Signed-off-by: Jens Axboe <[email protected]> commit 398205b8391b208f0034a392242867b28ad8af3d Author: Paolo Bonzini <[email protected]> Date: Fri Nov 7 23:03:59 2014 +0100 blk_mq: call preempt_disable/enable in blk_mq_run_hw_queue, and only if needed preempt_disable/enable surrounds every call to blk_mq_run_hw_queue, except the one in blk-flush.c. In fact that one is always asynchronous, and it does not need smp_processor_id(). We can do the same for all other calls, avoiding preempt_disable when async is true. This avoids peppering blk-mq.c with preemption-disabled regions. Cc: Jens Axboe <[email protected]> Cc: Thomas Gleixner <[email protected]> Reported-by: Clark Williams <[email protected]> Tested-by: Clark Williams <[email protected]> Signed-off-by: Paolo Bonzini <[email protected]> Signed-off-by: Jens Axboe <[email protected]> commit 9169051617df7fca597274e9e43324332cb8f0ee Author: Mark Brown <[email protected]> Date: Sat Nov 8 10:28:10 2014 +0000 spi: spidev: Don't mangle max_speed_hz in underlying spi device Currently spidev allows callers to set the default speed by overriding the max_speed_hz in the underlying device. This achieves the immediate goal but is not what devices expect and can easily lead to userspace trying to set unsupported speeds and succeeding, apart from anything else drivers can't set a limit on the speed using max_speed_hz as they'd expect and any other devices on the bus will be affected. Instead store the default speed in the spidev struct and fill this in on each transfer. Signed-off-by: Mark Brown <[email protected]> commit 2c8c56e15df3d4c2af3d656e44feb18789f75837 Author: Eric Dumazet <[email protected]> Date: Tue Nov 11 05:54:28 2014 -0800 net: introduce SO_INCOMING_CPU Alternative to RPS/RFS is to use hardware support for multiple queues. Then split a set of million of sockets into worker threads, each one using epoll() to manage events on its own socket pool. Ideally, we want one thread per RX/TX queue/cpu, but we have no way to know after accept() or connect() on which queue/cpu a socket is managed. We normally use one cpu per RX queue (IRQ smp_affinity being properly set), so remembering on socket structure which cpu delivered last packet is enough to solve the problem. After accept(), connect(), or even file descriptor passing around processes, applications can use : int cpu; socklen_t len = sizeof(cpu); getsockopt(fd, SOL_SOCKET, SO_INCOMING_CPU, &cpu, &len); And use this information to put the socket into the right silo for optimal performance, as all networking stack should run on the appropriate cpu, without need to send IPI (RPS/RFS). Signed-off-by: Eric Dumazet <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit 3d97379a67486bc481ab5b8f7aa5b7ceb6154a95 Author: Eric Dumazet <[email protected]> Date: Tue Nov 11 05:54:27 2014 -0800 tcp: move sk_mark_napi_id() at the right place sk_mark_napi_id() is used to record for a flow napi id of incoming packets for busypoll sake. We should do this only on established flows, not on listeners. This was 'working' by virtue of the socket cloning, but doing this on SYN packets in unecessary cache line dirtying. Even if we move sk_napi_id in the same cache line than sk_lock, we are working to make SYN processing lockless, so it is desirable to set sk_napi_id only for established flows. Signed-off-by: Eric Dumazet <[email protected]> Signed-off-by: David S. Miller <[email protected]> commit 26488b3723270880a28b542ff2276689506d6a9f Author: Jiang Liu <[email protected]> Date: Thu Aug 22 20:59:39 2013 +0800 tracing: Add entry->next_cpu to trace_ctxwake_bin() Function trace_ctxwake_bin() misses ctx_switch_entry->next_cpu field, so user will get stale value for "next_cpu". Link: http://lkml.kernel.org/p/[email protected] Signed-off-by: Jiang Liu <[email protected]> Signed-off-by: Steven Rostedt <[email protected]> commit 243f7610a68a606eb1787c09450a440bf30bebe0 Author: Steven Rostedt (Red Hat) <[email protected]> Date: Thu Oct 30 20:44:53 2014 -0400 tracing: Move tracing_sched_{switch,wakeup}() into wakeup tracer The only code that references tracing_sched_switch_trace() and tracing_sched_wakeup_trace() is the wakeup latency tracer. Those two functions use to belong to the sched_switch tracer which has long been removed. These functions were left behind because the wakeup latency tracer used them. But since the wakeup latency tracer is the only one to use them, they should be static functions inside that code. Signed-off-by: Steven Rostedt <[email protected]> commit 458faf0b88b19a46d51bb9760fa6e03a1bc6d97b Author: Oleg Nesterov <[email protected]> Date: Wed Jul 23 21:35:03 2014 +0200 tracing: Kill the dead code in probe_sched_switch() and probe_sched_wakeup() After the previous patch it is clear that "tracer_enabled" can never be true, we can remove the "if (tracer_enabled)" code in probe_sched_switch() and probe_sched_wakeup(). Plus we can obviously remove tracer_enabled, ctx_trace, and sched_stopped as well. Link: http://lkml.kernel.org/p/[email protected] Signed-off-by: Oleg Nesterov <[email protected]> Signed-off-by: Steven Rostedt <[email protected]> commit 632537256e9f969a188cc4d0159e0027a459d3e7 Author: Oleg Nesterov <[email protected]> Date: Wed Jul 23 21:35:01 2014 +0200 tracing: Kill tracing_{start,stop}_sched_switch_record() and tracing_sched_switch_assign_trace() tracing_{start,stop}_sched_switch_record() have no callers since 87d80de2800d "tracing: Remove obsolete sched_switch tracer". The last caller of tracing_sched_switch_assign_trace() was removed by 30dbb20e68e6 "tracing: Remove boot tracer". Link: http://lkml.kernel.org/p/[email protected] Signed-off-by: Oleg Nesterov <[email protected]> Signed-off-by: Steven Rostedt <[email protected]> commit 4fd3279b48605ae3ea509b9b2c02e46aa0975930 Author: Steven Rostedt (Red Hat) <[email protected]> Date: Fri Oct 24 17:56:04 2014 -0400 ftrace: Add more information to ftrace_bug() output With the introduction of the dynamic trampolines, it is useful that if things go wrong that ftrace_bug() produces more information about what the current state is. This can help debug issues that may arise. Ftrace has lots of checks to make sure that the state of the system it touchs is exactly what it expects it to be. When it detects an abnormality it calls ftrace_bug() and disables itself to prevent any further damage. It is crucial that ftrace_bug() produces sufficient information that can be used to debug the situation. Cc: Benjamin Herrenschmidt <[email protected]> Acked-by: Borislav Petkov <[email protected]> Tested-by: Masami Hiramatsu <[email protected]> Tested-by: Jiri Kosina <[email protected]> Signed-off-by: Steven Rostedt <[email protected]> commit 12cce594fa8f12e002e7eb5d10141853c1e6a112 Author: Steven Rostedt (Red Hat) <[email protected]> Date: Thu Jul 3 15:48:16 2014 -0400 ftrace/x86: Allow !CONFIG_PREEMPT dynamic ops to use allocated trampolines When the static ftrace_ops (like function tracer) enables tracing, and it is the only callback that is referencing a function, a trampoline is dynamically allocated to the function that calls the callback directly instead of calling a loop function that iterates over all the registered ftrace ops (if more than one ops is registered). But when it comes to dynamically allocated ftrace_ops, where they may be freed, on a CONFIG_PREEMPT kernel there's no way to know when it is safe to free the trampoline. If a task was preempted while executing on the trampoline, there's currently no way to know when it will be off that trampoline. But this is not true when it comes to !CONFIG_PREEMPT. The current method of calling schedule_on_each_cpu() will force tasks off the trampoline, becaues they can not schedule while on it (kernel preemption is not configured). That means it is safe to free a dynamically allocated ftrace ops trampoline when CONFIG_PREEMPT is not configured. Cc: H. Peter Anvin <[email protected]> Cc: Paul E. McKenney <[email protected]> Acked-by: Borislav Petkov <[email protected]> Tested-by: Masami Hiramatsu <[email protected]> Tested-by: Jiri Kosina <[email protected]> Signed-off-by: Steven Rostedt <[email protected]> commit 19ca5a3cc425cc9a8abedb0f4fb7b4e7ceee2255 Author: Andreas Ruprecht <[email protected]> Date: Sun Aug 10 21:10:03 2014 +0200 EDAC, pci_sysfs: remove unneccessary ifdef around entire file The file edac_pci_sysfs.c is dependent on CONFIG_PCI. This is already modelled in the Makefile, but edac_pci_sysfs.o is still contained in the list of files compiled even without CONFIG_PCI. This change removes edac_pci_sysfs.o from the list of built objects when not having CONFIG_PCI enabled and removes the then-unnecessary ifdef from the source file. Signed-off-by: Andreas Ruprecht <[email protected]> Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Borislav Petkov <[email protected]> commit 419a2ea074189be8de0b0ab052dd98061fed1c16 Author: Mike Snitzer <[email protected]> Date: Tue Oct 28 20:58:45 2014 -0400 dm thin: suspend/resume active thin devices when reloading thin-pool Before this change it was expected that userspace would first suspend all active thin devices, reload/resize the thin-pool target, then resume all active thin devices. Now the thin-pool suspend/resume will trigger the suspend/resume of all active thins via appropriate calls to dm_internal_suspend and dm_internal_resume. Store the mapped_device for each thin device in struct thin_c to make these calls possible. Signed-off-by: Mike Snitzer <[email protected]> commit 665aa8cdc499b9aeea6532e682a58ca34b7f94e6 Author: Dan Carpenter <[email protected]> Date: Fri Aug 1 11:25:14 2014 +0300 ghes_edac: Use snprintf() to silence a static checker warning My static checker complains because the "e->location" has up to 256 characters but we are copying it into the "pvt->detail_location" which only has space for 240 characters. That's not counting the surrounding text and the "e->other_detail" string which can be over 80 characters long. I am not familiar with this code but presumably it normally works. Let's add a limit though for safety. Signed-off-by: Dan Carpenter <[email protected]> Acked-by: Mauro Carvalho Chehab <[email protected]> Link: http://lkml.kernel.org/r/20140801082514.GD28869@mwanda Signed-off-by: Borislav Petkov <[email protected]> commit 8860704ea9afa699484c9fe7822da1cd37e40690 Author: Mike Snitzer <[email protected]> Date: Tue Oct 28 18:34:52 2014 -0400 dm: enhance internal suspend and resume interface Rename dm_internal_{suspend,resume} to dm_internal_{suspend,resume}_fast -- dm-stats will continue using these methods to avoid all the extra suspend/resume logic that is not needed in order to quickly flush IO. Introduce dm_internal_suspend_noflush() variant that actually calls the mapped_device's target callbacks -- otherwise target-specific hooks are avoided (e.g. dm-thin's thin_presuspend and thin_postsuspend). Common code between dm_internal_{suspend_noflush,resume} and dm_{suspend,resume} was factored out as __dm_{suspend,resume}. Update dm_internal_{suspend_noflush,resume} to always take and release the mapped_device's suspend_lock. Also update dm_{suspend,resume} to be aware of potential for DM_INTERNAL_SUSPEND_FLAG to be set and respond accordingly by interruptibly waiting for the DM_INTERNAL_SUSPEND_FLAG to be cleared. Add lockdep annotation to dm_suspend() and dm_resume(). Also add DM_INTERNAL_SUSPEND_FLAG to status report. This new DM_INTERNAL_SUSPEND_FLAG state is being tracked/reported to assist with debugging (e.g. 'dmsetup info' will report an internally suspended device accordingly). The existing DM_SUSPEND_FLAG remains unchanged. DM_INTERNAL_SUSPEND_FLAG is set by dm_internal_suspend_noflush() and cleared by dm_internal_resume(). Both DM_SUSPEND_FLAG and DM_INTERNAL_SUSPEND_FLAG may be set if a device was already suspended when dm_internal_suspend_noflush() was called -- this can be thought of as a "nested suspend". A "nested suspend" can with legacy userspace dm-thin code that might suspend all active thin volumes before suspending the pool for resize. But otherwise, in the normal dm-thin-pool suspend case moving forward: the thin-pool will have DM_SUSPEND_FLAG set and all active thins from that thin-pool will have DM_INTERNAL_SUSPEND_FLAG set. Signed-off-by: Mike Snitzer <[email protected]> commit bf735ebb898c1348f635488bee737b95886aa20e Author: Mike Snitzer <[email protected]> Date: Fri Nov 7 15:09:46 2014 -0500 dm thin: do not allow thin device activation while pool is suspended Otherwise IO could be issued to the pool while it is suspended. Care was taken to properly interlock between the thin and thin-pool targets when accessing the pool's 'suspended' flag. Signed-off-by: Mike Snitzer <[email protected]> commit 3ca4517dc53569ad82152ff035be0e909c06c722 Author: Mike Snitzer <[email protected]> Date: Tue Oct 28 20:13:31 2014 -0400 dm: add presuspend_undo hook to target_type The DM thin-pool target now must undo the changes performed during pool_presuspend() so introduce presuspend_undo hook in target_type. Signed-off-by: Mike Snitzer <[email protected]> commit f35d0679e47ef97c6037f7aefba41f18a7865421 Author: Mike Snitzer <[email protected]> Date: Fri Nov 7 15:27:56 2014 -0500 dm thin: remove stale 'trim' message in block comment above pool_message Signed-off-by: M…

…ion-cgu Ci20 v3.18 MSC0 Performance Fix

delete_object_part() can be called by multiple callers in the same time. If an object is found and removed by a caller, and then another caller try to find it too, it failed and return directly. It still be recorded by kmemleak even if it has already been freed to buddy. With DEBUG on, kmemleak will report the following warning, kmemleak: Partially freeing unknown object at 0xa1af86000 (size 4096) CPU: 0 PID: 742 Comm: test_huge Not tainted 6.6.0-rc3kmemleak+ torvalds#54 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.13.0-1ubuntu1.1 04/01/2014 Call Trace: <TASK> dump_stack_lvl+0x37/0x50 kmemleak_free_part_phys+0x50/0x60 hugetlb_vmemmap_optimize+0x172/0x290 ? __pfx_vmemmap_remap_pte+0x10/0x10 __prep_new_hugetlb_folio+0xe/0x30 prep_new_hugetlb_folio.isra.0+0xe/0x40 alloc_fresh_hugetlb_folio+0xc3/0xd0 alloc_surplus_hugetlb_folio.constprop.0+0x6e/0xd0 hugetlb_acct_memory.part.0+0xe6/0x2a0 hugetlb_reserve_pages+0x110/0x2c0 hugetlbfs_file_mmap+0x11d/0x1b0 mmap_region+0x248/0x9a0 ? hugetlb_get_unmapped_area+0x15c/0x2d0 do_mmap+0x38b/0x580 vm_mmap_pgoff+0xe6/0x190 ksys_mmap_pgoff+0x18a/0x1f0 do_syscall_64+0x3f/0x90 entry_SYSCALL_64_after_hwframe+0x6e/0xd8 Expand __create_object() and move __alloc_object() to the beginning. Then use kmemleak_lock to protect __find_and_remove_object() and __link_object() as a whole, which can guarantee all objects are processed sequentialally. Link: https://lkml.kernel.org/r/[email protected] Fixes: 53238a6 ("kmemleak: Allow partial freeing of memory blocks") Signed-off-by: Liu Shixin <[email protected]> Reviewed-by: Catalin Marinas <[email protected]> Cc: Kefeng Wang <[email protected]> Cc: Patrick Wang <[email protected]> Signed-off-by: Andrew Morton <[email protected]>

Bpf next merge

KSAN calls into rcu code which then triggers a write that reenters into KSAN getting the system stuck doing infinite recursion. #0 kmsan_get_context () at mm/kmsan/kmsan.h:106 #1 __msan_get_context_state () at mm/kmsan/instrumentation.c:331 #2 0xffffffff81495671 in get_current () at ./arch/x86/include/asm/current.h:42 #3 rcu_preempt_read_enter () at kernel/rcu/tree_plugin.h:379 #4 __rcu_read_lock () at kernel/rcu/tree_plugin.h:402 #5 0xffffffff81b2054b in rcu_read_lock () at ./include/linux/rcupdate.h:748 torvalds#6 pfn_valid (pfn=<optimized out>) at ./include/linux/mmzone.h:2016 torvalds#7 kmsan_virt_addr_valid (addr=addr@entry=0xffffffff8620d974 <init_task+1012>) at ./arch/x86/include/asm/kmsan.h:82 torvalds#8 virt_to_page_or_null (vaddr=vaddr@entry=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/shadow.c:75 torvalds#9 0xffffffff81b2023c in kmsan_get_metadata (address=0xffffffff8620d974 <init_task+1012>, is_origin=false) at mm/kmsan/shadow.c:143 torvalds#10 kmsan_get_shadow_origin_ptr (address=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/shadow.c:97 torvalds#11 0xffffffff81b1dbd2 in get_shadow_origin_ptr (addr=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/instrumentation.c:36 torvalds#12 __msan_metadata_ptr_for_load_4 (addr=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/instrumentation.c:91 torvalds#13 0xffffffff8149568f in rcu_preempt_read_enter () at kernel/rcu/tree_plugin.h:379 torvalds#14 __rcu_read_lock () at kernel/rcu/tree_plugin.h:402 torvalds#15 0xffffffff81b2054b in rcu_read_lock () at ./include/linux/rcupdate.h:748 torvalds#16 pfn_valid (pfn=<optimized out>) at ./include/linux/mmzone.h:2016 torvalds#17 kmsan_virt_addr_valid (addr=addr@entry=0xffffffff8620d974 <init_task+1012>) at ./arch/x86/include/asm/kmsan.h:82 torvalds#18 virt_to_page_or_null (vaddr=vaddr@entry=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/shadow.c:75 torvalds#19 0xffffffff81b2023c in kmsan_get_metadata (address=0xffffffff8620d974 <init_task+1012>, is_origin=false) at mm/kmsan/shadow.c:143 torvalds#20 kmsan_get_shadow_origin_ptr (address=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/shadow.c:97 torvalds#21 0xffffffff81b1dbd2 in get_shadow_origin_ptr (addr=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/instrumentation.c:36 torvalds#22 __msan_metadata_ptr_for_load_4 (addr=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/instrumentation.c:91 torvalds#23 0xffffffff8149568f in rcu_preempt_read_enter () at kernel/rcu/tree_plugin.h:379 torvalds#24 __rcu_read_lock () at kernel/rcu/tree_plugin.h:402 torvalds#25 0xffffffff81b2054b in rcu_read_lock () at ./include/linux/rcupdate.h:748 torvalds#26 pfn_valid (pfn=<optimized out>) at ./include/linux/mmzone.h:2016 torvalds#27 kmsan_virt_addr_valid (addr=addr@entry=0xffffffff8620d974 <init_task+1012>) at ./arch/x86/include/asm/kmsan.h:82 torvalds#28 virt_to_page_or_null (vaddr=vaddr@entry=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/shadow.c:75 torvalds#29 0xffffffff81b2023c in kmsan_get_metadata (address=0xffffffff8620d974 <init_task+1012>, is_origin=false) at mm/kmsan/shadow.c:143 torvalds#30 kmsan_get_shadow_origin_ptr (address=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/shadow.c:97 torvalds#31 0xffffffff81b1dbd2 in get_shadow_origin_ptr (addr=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/instrumentation.c:36 torvalds#32 __msan_metadata_ptr_for_load_4 (addr=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/instrumentation.c:91 torvalds#33 0xffffffff8149568f in rcu_preempt_read_enter () at kernel/rcu/tree_plugin.h:379 torvalds#34 __rcu_read_lock () at kernel/rcu/tree_plugin.h:402 torvalds#35 0xffffffff81b2054b in rcu_read_lock () at ./include/linux/rcupdate.h:748 torvalds#36 pfn_valid (pfn=<optimized out>) at ./include/linux/mmzone.h:2016 torvalds#37 kmsan_virt_addr_valid (addr=addr@entry=0xffffffff8620d974 <init_task+1012>) at ./arch/x86/include/asm/kmsan.h:82 torvalds#38 virt_to_page_or_null (vaddr=vaddr@entry=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/shadow.c:75 torvalds#39 0xffffffff81b2023c in kmsan_get_metadata (address=0xffffffff8620d974 <init_task+1012>, is_origin=false) at mm/kmsan/shadow.c:143 torvalds#40 kmsan_get_shadow_origin_ptr (address=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/shadow.c:97 torvalds#41 0xffffffff81b1dbd2 in get_shadow_origin_ptr (addr=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/instrumentation.c:36 torvalds#42 __msan_metadata_ptr_for_load_4 (addr=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/instrumentation.c:91 torvalds#43 0xffffffff8149568f in rcu_preempt_read_enter () at kernel/rcu/tree_plugin.h:379 torvalds#44 __rcu_read_lock () at kernel/rcu/tree_plugin.h:402 torvalds#45 0xffffffff81b2054b in rcu_read_lock () at ./include/linux/rcupdate.h:748 torvalds#46 pfn_valid (pfn=<optimized out>) at ./include/linux/mmzone.h:2016 torvalds#47 kmsan_virt_addr_valid (addr=addr@entry=0xffffffff8620d974 <init_task+1012>) at ./arch/x86/include/asm/kmsan.h:82 torvalds#48 virt_to_page_or_null (vaddr=vaddr@entry=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/shadow.c:75 torvalds#49 0xffffffff81b2023c in kmsan_get_metadata (address=0xffffffff8620d974 <init_task+1012>, is_origin=false) at mm/kmsan/shadow.c:143 torvalds#50 kmsan_get_shadow_origin_ptr (address=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/shadow.c:97 torvalds#51 0xffffffff81b1dbd2 in get_shadow_origin_ptr (addr=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/instrumentation.c:36 #52 __msan_metadata_ptr_for_load_4 (addr=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/instrumentation.c:91 #53 0xffffffff8149568f in rcu_preempt_read_enter () at kernel/rcu/tree_plugin.h:379 torvalds#54 __rcu_read_lock () at kernel/rcu/tree_plugin.h:402 torvalds#55 0xffffffff81b2054b in rcu_read_lock () at ./include/linux/rcupdate.h:748 torvalds#56 pfn_valid (pfn=<optimized out>) at ./include/linux/mmzone.h:2016 torvalds#57 kmsan_virt_addr_valid (addr=addr@entry=0xffffffff8620d974 <init_task+1012>) at ./arch/x86/include/asm/kmsan.h:82 #58 virt_to_page_or_null (vaddr=vaddr@entry=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/shadow.c:75 torvalds#59 0xffffffff81b2023c in kmsan_get_metadata (address=0xffffffff8620d974 <init_task+1012>, is_origin=false) at mm/kmsan/shadow.c:143 torvalds#60 kmsan_get_shadow_origin_ptr (address=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/shadow.c:97 torvalds#61 0xffffffff81b1dbd2 in get_shadow_origin_ptr (addr=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/instrumentation.c:36 torvalds#62 __msan_metadata_ptr_for_load_4 (addr=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/instrumentation.c:91 torvalds#63 0xffffffff8149568f in rcu_preempt_read_enter () at kernel/rcu/tree_plugin.h:379 torvalds#64 __rcu_read_lock () at kernel/rcu/tree_plugin.h:402 torvalds#65 0xffffffff81b2054b in rcu_read_lock () at ./include/linux/rcupdate.h:748 torvalds#66 pfn_valid (pfn=<optimized out>) at ./include/linux/mmzone.h:2016 torvalds#67 kmsan_virt_addr_valid (addr=addr@entry=0xffffffff8620d974 <init_task+1012>) at ./arch/x86/include/asm/kmsan.h:82 torvalds#68 virt_to_page_or_null (vaddr=vaddr@entry=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/shadow.c:75 torvalds#69 0xffffffff81b2023c in kmsan_get_metadata (address=0xffffffff8620d974 <init_task+1012>, is_origin=false) at mm/kmsan/shadow.c:143 #70 kmsan_get_shadow_origin_ptr (address=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/shadow.c:97 torvalds#71 0xffffffff81b1dbd2 in get_shadow_origin_ptr (addr=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/instrumentation.c:36 torvalds#72 __msan_metadata_ptr_for_load_4 (addr=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/instrumentation.c:91 torvalds#73 0xffffffff8149568f in rcu_preempt_read_enter () at kernel/rcu/tree_plugin.h:379 torvalds#74 __rcu_read_lock () at kernel/rcu/tree_plugin.h:402 torvalds#75 0xffffffff81b2054b in rcu_read_lock () at ./include/linux/rcupdate.h:748 torvalds#76 pfn_valid (pfn=<optimized out>) at ./include/linux/mmzone.h:2016 torvalds#77 kmsan_virt_addr_valid (addr=addr@entry=0xffffffff86203c90) at ./arch/x86/include/asm/kmsan.h:82 torvalds#78 virt_to_page_or_null (vaddr=vaddr@entry=0xffffffff86203c90) at mm/kmsan/shadow.c:75 torvalds#79 0xffffffff81b2023c in kmsan_get_metadata (address=0xffffffff86203c90, is_origin=false) at mm/kmsan/shadow.c:143 torvalds#80 kmsan_get_shadow_origin_ptr (address=0xffffffff86203c90, size=8, store=false) at mm/kmsan/shadow.c:97 torvalds#81 0xffffffff81b1dc72 in get_shadow_origin_ptr (addr=0xffffffff8620d974 <init_task+1012>, size=8, store=false) at mm/kmsan/instrumentation.c:36 torvalds#82 __msan_metadata_ptr_for_load_8 (addr=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/instrumentation.c:92 torvalds#83 0xffffffff814fdb9e in filter_irq_stacks (entries=<optimized out>, nr_entries=4) at kernel/stacktrace.c:397 torvalds#84 0xffffffff829520e8 in stack_depot_save_flags (entries=0xffffffff8620d974 <init_task+1012>, nr_entries=4, alloc_flags=0, depot_flags=0) at lib/stackdepot.c:500 torvalds#85 0xffffffff81b1e560 in __msan_poison_alloca (address=0xffffffff86203da0, size=24, descr=<optimized out>) at mm/kmsan/instrumentation.c:285 torvalds#86 0xffffffff8562821c in _printk (fmt=0xffffffff85f191a5 "\0016Attempting lock1") at kernel/printk/printk.c:2324 torvalds#87 0xffffffff81942aa2 in kmem_cache_create_usercopy (name=0xffffffff85f18903 "mm_struct", size=1296, align=0, flags=270336, useroffset=<optimized out>, usersize=<optimized out>, ctor=0x0 <fixed_percpu_data>) at mm/slab_common.c:296 torvalds#88 0xffffffff86f337a0 in mm_cache_init () at kernel/fork.c:3262 torvalds#89 0xffffffff86eacb8e in start_kernel () at init/main.c:932 torvalds#90 0xffffffff86ecdf94 in x86_64_start_reservations (real_mode_data=0x140e0 <exception_stacks+28896> <error: Cannot access memory at address 0x140e0>) at arch/x86/kernel/head64.c:555 torvalds#91 0xffffffff86ecde9b in x86_64_start_kernel (real_mode_data=0x140e0 <exception_stacks+28896> <error: Cannot access memory at address 0x140e0>) at arch/x86/kernel/head64.c:536 torvalds#92 0xffffffff810001d3 in secondary_startup_64 () at /pool/workspace/linux/arch/x86/kernel/head_64.S:461 torvalds#93 0x0000000000000000 in ??

As of 5ec8e8e(mm/sparsemem: fix race in accessing memory_section->usage) KMSAN now calls into RCU tree code during kmsan_get_metadata. This will trigger a write that will reenter into KMSAN getting the system stuck doing infinite recursion. #0 kmsan_get_context () at mm/kmsan/kmsan.h:106 #1 __msan_get_context_state () at mm/kmsan/instrumentation.c:331 #2 0xffffffff81495671 in get_current () at ./arch/x86/include/asm/current.h:42 #3 rcu_preempt_read_enter () at kernel/rcu/tree_plugin.h:379 #4 __rcu_read_lock () at kernel/rcu/tree_plugin.h:402 #5 0xffffffff81b2054b in rcu_read_lock () at ./include/linux/rcupdate.h:748 torvalds#6 pfn_valid (pfn=<optimized out>) at ./include/linux/mmzone.h:2016 torvalds#7 kmsan_virt_addr_valid (addr=addr@entry=0xffffffff8620d974 <init_task+1012>) at ./arch/x86/include/asm/kmsan.h:82 torvalds#8 virt_to_page_or_null (vaddr=vaddr@entry=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/shadow.c:75 torvalds#9 0xffffffff81b2023c in kmsan_get_metadata (address=0xffffffff8620d974 <init_task+1012>, is_origin=false) at mm/kmsan/shadow.c:143 torvalds#10 kmsan_get_shadow_origin_ptr (address=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/shadow.c:97 torvalds#11 0xffffffff81b1dbd2 in get_shadow_origin_ptr (addr=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/instrumentation.c:36 torvalds#12 __msan_metadata_ptr_for_load_4 (addr=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/instrumentation.c:91 torvalds#13 0xffffffff8149568f in rcu_preempt_read_enter () at kernel/rcu/tree_plugin.h:379 torvalds#14 __rcu_read_lock () at kernel/rcu/tree_plugin.h:402 torvalds#15 0xffffffff81b2054b in rcu_read_lock () at ./include/linux/rcupdate.h:748 torvalds#16 pfn_valid (pfn=<optimized out>) at ./include/linux/mmzone.h:2016 torvalds#17 kmsan_virt_addr_valid (addr=addr@entry=0xffffffff8620d974 <init_task+1012>) at ./arch/x86/include/asm/kmsan.h:82 torvalds#18 virt_to_page_or_null (vaddr=vaddr@entry=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/shadow.c:75 torvalds#19 0xffffffff81b2023c in kmsan_get_metadata (address=0xffffffff8620d974 <init_task+1012>, is_origin=false) at mm/kmsan/shadow.c:143 torvalds#20 kmsan_get_shadow_origin_ptr (address=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/shadow.c:97 torvalds#21 0xffffffff81b1dbd2 in get_shadow_origin_ptr (addr=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/instrumentation.c:36 torvalds#22 __msan_metadata_ptr_for_load_4 (addr=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/instrumentation.c:91 torvalds#23 0xffffffff8149568f in rcu_preempt_read_enter () at kernel/rcu/tree_plugin.h:379 torvalds#24 __rcu_read_lock () at kernel/rcu/tree_plugin.h:402 torvalds#25 0xffffffff81b2054b in rcu_read_lock () at ./include/linux/rcupdate.h:748 torvalds#26 pfn_valid (pfn=<optimized out>) at ./include/linux/mmzone.h:2016 torvalds#27 kmsan_virt_addr_valid (addr=addr@entry=0xffffffff8620d974 <init_task+1012>) at ./arch/x86/include/asm/kmsan.h:82 torvalds#28 virt_to_page_or_null (vaddr=vaddr@entry=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/shadow.c:75 torvalds#29 0xffffffff81b2023c in kmsan_get_metadata (address=0xffffffff8620d974 <init_task+1012>, is_origin=false) at mm/kmsan/shadow.c:143 torvalds#30 kmsan_get_shadow_origin_ptr (address=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/shadow.c:97 torvalds#31 0xffffffff81b1dbd2 in get_shadow_origin_ptr (addr=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/instrumentation.c:36 torvalds#32 __msan_metadata_ptr_for_load_4 (addr=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/instrumentation.c:91 torvalds#33 0xffffffff8149568f in rcu_preempt_read_enter () at kernel/rcu/tree_plugin.h:379 torvalds#34 __rcu_read_lock () at kernel/rcu/tree_plugin.h:402 torvalds#35 0xffffffff81b2054b in rcu_read_lock () at ./include/linux/rcupdate.h:748 torvalds#36 pfn_valid (pfn=<optimized out>) at ./include/linux/mmzone.h:2016 torvalds#37 kmsan_virt_addr_valid (addr=addr@entry=0xffffffff8620d974 <init_task+1012>) at ./arch/x86/include/asm/kmsan.h:82 torvalds#38 virt_to_page_or_null (vaddr=vaddr@entry=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/shadow.c:75 torvalds#39 0xffffffff81b2023c in kmsan_get_metadata (address=0xffffffff8620d974 <init_task+1012>, is_origin=false) at mm/kmsan/shadow.c:143 torvalds#40 kmsan_get_shadow_origin_ptr (address=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/shadow.c:97 torvalds#41 0xffffffff81b1dbd2 in get_shadow_origin_ptr (addr=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/instrumentation.c:36 torvalds#42 __msan_metadata_ptr_for_load_4 (addr=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/instrumentation.c:91 torvalds#43 0xffffffff8149568f in rcu_preempt_read_enter () at kernel/rcu/tree_plugin.h:379 torvalds#44 __rcu_read_lock () at kernel/rcu/tree_plugin.h:402 torvalds#45 0xffffffff81b2054b in rcu_read_lock () at ./include/linux/rcupdate.h:748 torvalds#46 pfn_valid (pfn=<optimized out>) at ./include/linux/mmzone.h:2016 torvalds#47 kmsan_virt_addr_valid (addr=addr@entry=0xffffffff8620d974 <init_task+1012>) at ./arch/x86/include/asm/kmsan.h:82 torvalds#48 virt_to_page_or_null (vaddr=vaddr@entry=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/shadow.c:75 torvalds#49 0xffffffff81b2023c in kmsan_get_metadata (address=0xffffffff8620d974 <init_task+1012>, is_origin=false) at mm/kmsan/shadow.c:143 torvalds#50 kmsan_get_shadow_origin_ptr (address=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/shadow.c:97 torvalds#51 0xffffffff81b1dbd2 in get_shadow_origin_ptr (addr=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/instrumentation.c:36 #52 __msan_metadata_ptr_for_load_4 (addr=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/instrumentation.c:91 #53 0xffffffff8149568f in rcu_preempt_read_enter () at kernel/rcu/tree_plugin.h:379 torvalds#54 __rcu_read_lock () at kernel/rcu/tree_plugin.h:402 torvalds#55 0xffffffff81b2054b in rcu_read_lock () at ./include/linux/rcupdate.h:748 torvalds#56 pfn_valid (pfn=<optimized out>) at ./include/linux/mmzone.h:2016 torvalds#57 kmsan_virt_addr_valid (addr=addr@entry=0xffffffff8620d974 <init_task+1012>) at ./arch/x86/include/asm/kmsan.h:82 #58 virt_to_page_or_null (vaddr=vaddr@entry=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/shadow.c:75 torvalds#59 0xffffffff81b2023c in kmsan_get_metadata (address=0xffffffff8620d974 <init_task+1012>, is_origin=false) at mm/kmsan/shadow.c:143 torvalds#60 kmsan_get_shadow_origin_ptr (address=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/shadow.c:97 torvalds#61 0xffffffff81b1dbd2 in get_shadow_origin_ptr (addr=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/instrumentation.c:36 torvalds#62 __msan_metadata_ptr_for_load_4 (addr=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/instrumentation.c:91 torvalds#63 0xffffffff8149568f in rcu_preempt_read_enter () at kernel/rcu/tree_plugin.h:379 torvalds#64 __rcu_read_lock () at kernel/rcu/tree_plugin.h:402 torvalds#65 0xffffffff81b2054b in rcu_read_lock () at ./include/linux/rcupdate.h:748 torvalds#66 pfn_valid (pfn=<optimized out>) at ./include/linux/mmzone.h:2016 torvalds#67 kmsan_virt_addr_valid (addr=addr@entry=0xffffffff8620d974 <init_task+1012>) at ./arch/x86/include/asm/kmsan.h:82 torvalds#68 virt_to_page_or_null (vaddr=vaddr@entry=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/shadow.c:75 torvalds#69 0xffffffff81b2023c in kmsan_get_metadata (address=0xffffffff8620d974 <init_task+1012>, is_origin=false) at mm/kmsan/shadow.c:143 #70 kmsan_get_shadow_origin_ptr (address=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/shadow.c:97 torvalds#71 0xffffffff81b1dbd2 in get_shadow_origin_ptr (addr=0xffffffff8620d974 <init_task+1012>, size=4, store=false) at mm/kmsan/instrumentation.c:36 torvalds#72 __msan_metadata_ptr_for_load_4 (addr=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/instrumentation.c:91 torvalds#73 0xffffffff8149568f in rcu_preempt_read_enter () at kernel/rcu/tree_plugin.h:379 torvalds#74 __rcu_read_lock () at kernel/rcu/tree_plugin.h:402 torvalds#75 0xffffffff81b2054b in rcu_read_lock () at ./include/linux/rcupdate.h:748 torvalds#76 pfn_valid (pfn=<optimized out>) at ./include/linux/mmzone.h:2016 torvalds#77 kmsan_virt_addr_valid (addr=addr@entry=0xffffffff86203c90) at ./arch/x86/include/asm/kmsan.h:82 torvalds#78 virt_to_page_or_null (vaddr=vaddr@entry=0xffffffff86203c90) at mm/kmsan/shadow.c:75 torvalds#79 0xffffffff81b2023c in kmsan_get_metadata (address=0xffffffff86203c90, is_origin=false) at mm/kmsan/shadow.c:143 torvalds#80 kmsan_get_shadow_origin_ptr (address=0xffffffff86203c90, size=8, store=false) at mm/kmsan/shadow.c:97 torvalds#81 0xffffffff81b1dc72 in get_shadow_origin_ptr (addr=0xffffffff8620d974 <init_task+1012>, size=8, store=false) at mm/kmsan/instrumentation.c:36 torvalds#82 __msan_metadata_ptr_for_load_8 (addr=0xffffffff8620d974 <init_task+1012>) at mm/kmsan/instrumentation.c:92 torvalds#83 0xffffffff814fdb9e in filter_irq_stacks (entries=<optimized out>, nr_entries=4) at kernel/stacktrace.c:397 torvalds#84 0xffffffff829520e8 in stack_depot_save_flags (entries=0xffffffff8620d974 <init_task+1012>, nr_entries=4, alloc_flags=0, depot_flags=0) at lib/stackdepot.c:500 torvalds#85 0xffffffff81b1e560 in __msan_poison_alloca (address=0xffffffff86203da0, size=24, descr=<optimized out>) at mm/kmsan/instrumentation.c:285 torvalds#86 0xffffffff8562821c in _printk (fmt=0xffffffff85f191a5 "\0016Attempting lock1") at kernel/printk/printk.c:2324 torvalds#87 0xffffffff81942aa2 in kmem_cache_create_usercopy (name=0xffffffff85f18903 "mm_struct", size=1296, align=0, flags=270336, useroffset=<optimized out>, usersize=<optimized out>, ctor=0x0 <fixed_percpu_data>) at mm/slab_common.c:296 torvalds#88 0xffffffff86f337a0 in mm_cache_init () at kernel/fork.c:3262 torvalds#89 0xffffffff86eacb8e in start_kernel () at init/main.c:932 torvalds#90 0xffffffff86ecdf94 in x86_64_start_reservations (real_mode_data=0x140e0 <exception_stacks+28896> <error: Cannot access memory at address 0x140e0>) at arch/x86/kernel/head64.c:555 torvalds#91 0xffffffff86ecde9b in x86_64_start_kernel (real_mode_data=0x140e0 <exception_stacks+28896> <error: Cannot access memory at address 0x140e0>) at arch/x86/kernel/head64.c:536 torvalds#92 0xffffffff810001d3 in secondary_startup_64 () at /pool/workspace/linux/arch/x86/kernel/head_64.S:461 torvalds#93 0x0000000000000000 in ??

9f74a3d ("ice: Fix VF Reset paths when interface in a failed over aggregate"), the ice driver has acquired the LAG mutex in ice_reset_vf(). The commit placed this lock acquisition just prior to the acquisition of the VF configuration lock. If ice_reset_vf() acquires the configuration lock via the ICE_VF_RESET_LOCK flag, this could deadlock with ice_vc_cfg_qs_msg() because it always acquires the locks in the order of the VF configuration lock and then the LAG mutex. Lockdep reports this violation almost immediately on creating and then removing 2 VF: ====================================================== WARNING: possible circular locking dependency detected 6.8.0-rc6 torvalds#54 Tainted: G W O ------------------------------------------------------ kworker/60:3/6771 is trying to acquire lock: ff40d43e099380a0 (&vf->cfg_lock){+.+.}-{3:3}, at: ice_reset_vf+0x22f/0x4d0 [ice] but task is already holding lock: ff40d43ea1961210 (&pf->lag_mutex){+.+.}-{3:3}, at: ice_reset_vf+0xb7/0x4d0 [ice] which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #1 (&pf->lag_mutex){+.+.}-{3:3}: __lock_acquire+0x4f8/0xb40 lock_acquire+0xd4/0x2d0 __mutex_lock+0x9b/0xbf0 ice_vc_cfg_qs_msg+0x45/0x690 [ice] ice_vc_process_vf_msg+0x4f5/0x870 [ice] __ice_clean_ctrlq+0x2b5/0x600 [ice] ice_service_task+0x2c9/0x480 [ice] process_one_work+0x1e9/0x4d0 worker_thread+0x1e1/0x3d0 kthread+0x104/0x140 ret_from_fork+0x31/0x50 ret_from_fork_asm+0x1b/0x30 -> #0 (&vf->cfg_lock){+.+.}-{3:3}: check_prev_add+0xe2/0xc50 validate_chain+0x558/0x800 __lock_acquire+0x4f8/0xb40 lock_acquire+0xd4/0x2d0 __mutex_lock+0x9b/0xbf0 ice_reset_vf+0x22f/0x4d0 [ice] ice_process_vflr_event+0x98/0xd0 [ice] ice_service_task+0x1cc/0x480 [ice] process_one_work+0x1e9/0x4d0 worker_thread+0x1e1/0x3d0 kthread+0x104/0x140 ret_from_fork+0x31/0x50 ret_from_fork_asm+0x1b/0x30 other info that might help us debug this: Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(&pf->lag_mutex); lock(&vf->cfg_lock); lock(&pf->lag_mutex); lock(&vf->cfg_lock); *** DEADLOCK *** 4 locks held by kworker/60:3/6771: #0: ff40d43e05428b38 ((wq_completion)ice){+.+.}-{0:0}, at: process_one_work+0x176/0x4d0 #1: ff50d06e05197e58 ((work_completion)(&pf->serv_task)){+.+.}-{0:0}, at: process_one_work+0x176/0x4d0 #2: ff40d43ea1960e50 (&pf->vfs.table_lock){+.+.}-{3:3}, at: ice_process_vflr_event+0x48/0xd0 [ice] #3: ff40d43ea1961210 (&pf->lag_mutex){+.+.}-{3:3}, at: ice_reset_vf+0xb7/0x4d0 [ice] stack backtrace: CPU: 60 PID: 6771 Comm: kworker/60:3 Tainted: G W O 6.8.0-rc6 torvalds#54 Hardware name: Workqueue: ice ice_service_task [ice] Call Trace: <TASK> dump_stack_lvl+0x4a/0x80 check_noncircular+0x12d/0x150 check_prev_add+0xe2/0xc50 ? save_trace+0x59/0x230 ? add_chain_cache+0x109/0x450 validate_chain+0x558/0x800 __lock_acquire+0x4f8/0xb40 ? lockdep_hardirqs_on+0x7d/0x100 lock_acquire+0xd4/0x2d0 ? ice_reset_vf+0x22f/0x4d0 [ice] ? lock_is_held_type+0xc7/0x120 __mutex_lock+0x9b/0xbf0 ? ice_reset_vf+0x22f/0x4d0 [ice] ? ice_reset_vf+0x22f/0x4d0 [ice] ? rcu_is_watching+0x11/0x50 ? ice_reset_vf+0x22f/0x4d0 [ice] ice_reset_vf+0x22f/0x4d0 [ice] ? process_one_work+0x176/0x4d0 ice_process_vflr_event+0x98/0xd0 [ice] ice_service_task+0x1cc/0x480 [ice] process_one_work+0x1e9/0x4d0 worker_thread+0x1e1/0x3d0 ? __pfx_worker_thread+0x10/0x10 kthread+0x104/0x140 ? __pfx_kthread+0x10/0x10 ret_from_fork+0x31/0x50 ? __pfx_kthread+0x10/0x10 ret_from_fork_asm+0x1b/0x30 </TASK> To avoid deadlock, we must acquire the LAG mutex only after acquiring the VF configuration lock. Fix the ice_reset_vf() to acquire the LAG mutex only after we either acquire or check that the VF configuration lock is held. Fixes: 9f74a3d ("ice: Fix VF Reset paths when interface in a failed over aggregate") Signed-off-by: Jacob Keller <[email protected]> Reviewed-by: Dave Ertman <[email protected]> Reviewed-by: Mateusz Polchlopek <[email protected]> Tested-by: Przemek Kitszel <[email protected]> Tested-by: Rafal Romanowski <[email protected]> Signed-off-by: Tony Nguyen <[email protected]> Signed-off-by: NipaLocal <nipa@local>

9f74a3d ("ice: Fix VF Reset paths when interface in a failed over aggregate"), the ice driver has acquired the LAG mutex in ice_reset_vf(). The commit placed this lock acquisition just prior to the acquisition of the VF configuration lock. If ice_reset_vf() acquires the configuration lock via the ICE_VF_RESET_LOCK flag, this could deadlock with ice_vc_cfg_qs_msg() because it always acquires the locks in the order of the VF configuration lock and then the LAG mutex. Lockdep reports this violation almost immediately on creating and then removing 2 VF: ====================================================== WARNING: possible circular locking dependency detected 6.8.0-rc6 #54 Tainted: G W O ------------------------------------------------------ kworker/60:3/6771 is trying to acquire lock: ff40d43e099380a0 (&vf->cfg_lock){+.+.}-{3:3}, at: ice_reset_vf+0x22f/0x4d0 [ice] but task is already holding lock: ff40d43ea1961210 (&pf->lag_mutex){+.+.}-{3:3}, at: ice_reset_vf+0xb7/0x4d0 [ice] which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #1 (&pf->lag_mutex){+.+.}-{3:3}: __lock_acquire+0x4f8/0xb40 lock_acquire+0xd4/0x2d0 __mutex_lock+0x9b/0xbf0 ice_vc_cfg_qs_msg+0x45/0x690 [ice] ice_vc_process_vf_msg+0x4f5/0x870 [ice] __ice_clean_ctrlq+0x2b5/0x600 [ice] ice_service_task+0x2c9/0x480 [ice] process_one_work+0x1e9/0x4d0 worker_thread+0x1e1/0x3d0 kthread+0x104/0x140 ret_from_fork+0x31/0x50 ret_from_fork_asm+0x1b/0x30 -> #0 (&vf->cfg_lock){+.+.}-{3:3}: check_prev_add+0xe2/0xc50 validate_chain+0x558/0x800 __lock_acquire+0x4f8/0xb40 lock_acquire+0xd4/0x2d0 __mutex_lock+0x9b/0xbf0 ice_reset_vf+0x22f/0x4d0 [ice] ice_process_vflr_event+0x98/0xd0 [ice] ice_service_task+0x1cc/0x480 [ice] process_one_work+0x1e9/0x4d0 worker_thread+0x1e1/0x3d0 kthread+0x104/0x140 ret_from_fork+0x31/0x50 ret_from_fork_asm+0x1b/0x30 other info that might help us debug this: Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(&pf->lag_mutex); lock(&vf->cfg_lock); lock(&pf->lag_mutex); lock(&vf->cfg_lock); *** DEADLOCK *** 4 locks held by kworker/60:3/6771: #0: ff40d43e05428b38 ((wq_completion)ice){+.+.}-{0:0}, at: process_one_work+0x176/0x4d0 #1: ff50d06e05197e58 ((work_completion)(&pf->serv_task)){+.+.}-{0:0}, at: process_one_work+0x176/0x4d0 #2: ff40d43ea1960e50 (&pf->vfs.table_lock){+.+.}-{3:3}, at: ice_process_vflr_event+0x48/0xd0 [ice] #3: ff40d43ea1961210 (&pf->lag_mutex){+.+.}-{3:3}, at: ice_reset_vf+0xb7/0x4d0 [ice] stack backtrace: CPU: 60 PID: 6771 Comm: kworker/60:3 Tainted: G W O 6.8.0-rc6 #54 Hardware name: Workqueue: ice ice_service_task [ice] Call Trace: <TASK> dump_stack_lvl+0x4a/0x80 check_noncircular+0x12d/0x150 check_prev_add+0xe2/0xc50 ? save_trace+0x59/0x230 ? add_chain_cache+0x109/0x450 validate_chain+0x558/0x800 __lock_acquire+0x4f8/0xb40 ? lockdep_hardirqs_on+0x7d/0x100 lock_acquire+0xd4/0x2d0 ? ice_reset_vf+0x22f/0x4d0 [ice] ? lock_is_held_type+0xc7/0x120 __mutex_lock+0x9b/0xbf0 ? ice_reset_vf+0x22f/0x4d0 [ice] ? ice_reset_vf+0x22f/0x4d0 [ice] ? rcu_is_watching+0x11/0x50 ? ice_reset_vf+0x22f/0x4d0 [ice] ice_reset_vf+0x22f/0x4d0 [ice] ? process_one_work+0x176/0x4d0 ice_process_vflr_event+0x98/0xd0 [ice] ice_service_task+0x1cc/0x480 [ice] process_one_work+0x1e9/0x4d0 worker_thread+0x1e1/0x3d0 ? __pfx_worker_thread+0x10/0x10 kthread+0x104/0x140 ? __pfx_kthread+0x10/0x10 ret_from_fork+0x31/0x50 ? __pfx_kthread+0x10/0x10 ret_from_fork_asm+0x1b/0x30 </TASK> To avoid deadlock, we must acquire the LAG mutex only after acquiring the VF configuration lock. Fix the ice_reset_vf() to acquire the LAG mutex only after we either acquire or check that the VF configuration lock is held. Fixes: 9f74a3d ("ice: Fix VF Reset paths when interface in a failed over aggregate") Signed-off-by: Jacob Keller <[email protected]> Reviewed-by: Dave Ertman <[email protected]> Reviewed-by: Mateusz Polchlopek <[email protected]> Tested-by: Przemek Kitszel <[email protected]> Tested-by: Rafal Romanowski <[email protected]> Signed-off-by: Tony Nguyen <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]>

[ Upstream commit 96fdd1f ] 9f74a3d ("ice: Fix VF Reset paths when interface in a failed over aggregate"), the ice driver has acquired the LAG mutex in ice_reset_vf(). The commit placed this lock acquisition just prior to the acquisition of the VF configuration lock. If ice_reset_vf() acquires the configuration lock via the ICE_VF_RESET_LOCK flag, this could deadlock with ice_vc_cfg_qs_msg() because it always acquires the locks in the order of the VF configuration lock and then the LAG mutex. Lockdep reports this violation almost immediately on creating and then removing 2 VF: ====================================================== WARNING: possible circular locking dependency detected 6.8.0-rc6 torvalds#54 Tainted: G W O ------------------------------------------------------ kworker/60:3/6771 is trying to acquire lock: ff40d43e099380a0 (&vf->cfg_lock){+.+.}-{3:3}, at: ice_reset_vf+0x22f/0x4d0 [ice] but task is already holding lock: ff40d43ea1961210 (&pf->lag_mutex){+.+.}-{3:3}, at: ice_reset_vf+0xb7/0x4d0 [ice] which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #1 (&pf->lag_mutex){+.+.}-{3:3}: __lock_acquire+0x4f8/0xb40 lock_acquire+0xd4/0x2d0 __mutex_lock+0x9b/0xbf0 ice_vc_cfg_qs_msg+0x45/0x690 [ice] ice_vc_process_vf_msg+0x4f5/0x870 [ice] __ice_clean_ctrlq+0x2b5/0x600 [ice] ice_service_task+0x2c9/0x480 [ice] process_one_work+0x1e9/0x4d0 worker_thread+0x1e1/0x3d0 kthread+0x104/0x140 ret_from_fork+0x31/0x50 ret_from_fork_asm+0x1b/0x30 -> #0 (&vf->cfg_lock){+.+.}-{3:3}: check_prev_add+0xe2/0xc50 validate_chain+0x558/0x800 __lock_acquire+0x4f8/0xb40 lock_acquire+0xd4/0x2d0 __mutex_lock+0x9b/0xbf0 ice_reset_vf+0x22f/0x4d0 [ice] ice_process_vflr_event+0x98/0xd0 [ice] ice_service_task+0x1cc/0x480 [ice] process_one_work+0x1e9/0x4d0 worker_thread+0x1e1/0x3d0 kthread+0x104/0x140 ret_from_fork+0x31/0x50 ret_from_fork_asm+0x1b/0x30 other info that might help us debug this: Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(&pf->lag_mutex); lock(&vf->cfg_lock); lock(&pf->lag_mutex); lock(&vf->cfg_lock); *** DEADLOCK *** 4 locks held by kworker/60:3/6771: #0: ff40d43e05428b38 ((wq_completion)ice){+.+.}-{0:0}, at: process_one_work+0x176/0x4d0 #1: ff50d06e05197e58 ((work_completion)(&pf->serv_task)){+.+.}-{0:0}, at: process_one_work+0x176/0x4d0 #2: ff40d43ea1960e50 (&pf->vfs.table_lock){+.+.}-{3:3}, at: ice_process_vflr_event+0x48/0xd0 [ice] #3: ff40d43ea1961210 (&pf->lag_mutex){+.+.}-{3:3}, at: ice_reset_vf+0xb7/0x4d0 [ice] stack backtrace: CPU: 60 PID: 6771 Comm: kworker/60:3 Tainted: G W O 6.8.0-rc6 torvalds#54 Hardware name: Workqueue: ice ice_service_task [ice] Call Trace: <TASK> dump_stack_lvl+0x4a/0x80 check_noncircular+0x12d/0x150 check_prev_add+0xe2/0xc50 ? save_trace+0x59/0x230 ? add_chain_cache+0x109/0x450 validate_chain+0x558/0x800 __lock_acquire+0x4f8/0xb40 ? lockdep_hardirqs_on+0x7d/0x100 lock_acquire+0xd4/0x2d0 ? ice_reset_vf+0x22f/0x4d0 [ice] ? lock_is_held_type+0xc7/0x120 __mutex_lock+0x9b/0xbf0 ? ice_reset_vf+0x22f/0x4d0 [ice] ? ice_reset_vf+0x22f/0x4d0 [ice] ? rcu_is_watching+0x11/0x50 ? ice_reset_vf+0x22f/0x4d0 [ice] ice_reset_vf+0x22f/0x4d0 [ice] ? process_one_work+0x176/0x4d0 ice_process_vflr_event+0x98/0xd0 [ice] ice_service_task+0x1cc/0x480 [ice] process_one_work+0x1e9/0x4d0 worker_thread+0x1e1/0x3d0 ? __pfx_worker_thread+0x10/0x10 kthread+0x104/0x140 ? __pfx_kthread+0x10/0x10 ret_from_fork+0x31/0x50 ? __pfx_kthread+0x10/0x10 ret_from_fork_asm+0x1b/0x30 </TASK> To avoid deadlock, we must acquire the LAG mutex only after acquiring the VF configuration lock. Fix the ice_reset_vf() to acquire the LAG mutex only after we either acquire or check that the VF configuration lock is held. Fixes: 9f74a3d ("ice: Fix VF Reset paths when interface in a failed over aggregate") Signed-off-by: Jacob Keller <[email protected]> Reviewed-by: Dave Ertman <[email protected]> Reviewed-by: Mateusz Polchlopek <[email protected]> Tested-by: Przemek Kitszel <[email protected]> Tested-by: Rafal Romanowski <[email protected]> Signed-off-by: Tony Nguyen <[email protected]> Link: https://lore.kernel.org/r/[email protected] Signed-off-by: Jakub Kicinski <[email protected]> Signed-off-by: Sasha Levin <[email protected]>

use-after-free error in lock_sock_nested is reported: [ 179.140137][ T3731] ===================================================== [ 179.142675][ T3731] BUG: KMSAN: use-after-free in lock_sock_nested+0x280/0x2c0 [ 179.145494][ T3731] CPU: 4 PID: 3731 Comm: kworker/4:2 Not tainted 5.12.0-rc6+ torvalds#54 [ 179.148432][ T3731] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.13.0-1ubuntu1.1 04/01/2014 [ 179.151806][ T3731] Workqueue: events l2cap_chan_timeout [ 179.152730][ T3731] Call Trace: [ 179.153301][ T3731] dump_stack+0x24c/0x2e0 [ 179.154063][ T3731] kmsan_report+0xfb/0x1e0 [ 179.154855][ T3731] __msan_warning+0x5c/0xa0 [ 179.155579][ T3731] lock_sock_nested+0x280/0x2c0 [ 179.156436][ T3731] ? kmsan_get_metadata+0x116/0x180 [ 179.157257][ T3731] l2cap_sock_teardown_cb+0xb8/0x890 [ 179.158154][ T3731] ? __msan_metadata_ptr_for_load_8+0x10/0x20 [ 179.159141][ T3731] ? kmsan_get_metadata+0x116/0x180 [ 179.159994][ T3731] ? kmsan_get_shadow_origin_ptr+0x84/0xb0 [ 179.160959][ T3731] ? l2cap_sock_recv_cb+0x420/0x420 [ 179.161834][ T3731] l2cap_chan_del+0x3e1/0x1d50 [ 179.162608][ T3731] ? kmsan_get_metadata+0x116/0x180 [ 179.163435][ T3731] ? kmsan_get_shadow_origin_ptr+0x84/0xb0 [ 179.164406][ T3731] l2cap_chan_close+0xeea/0x1050 [ 179.165189][ T3731] ? kmsan_internal_unpoison_shadow+0x42/0x70 [ 179.166180][ T3731] l2cap_chan_timeout+0x1da/0x590 [ 179.167066][ T3731] ? __msan_metadata_ptr_for_load_8+0x10/0x20 [ 179.168023][ T3731] ? l2cap_chan_create+0x560/0x560 [ 179.168818][ T3731] process_one_work+0x121d/0x1ff0 [ 179.169598][ T3731] worker_thread+0x121b/0x2370 [ 179.170346][ T3731] kthread+0x4ef/0x610 [ 179.171010][ T3731] ? process_one_work+0x1ff0/0x1ff0 [ 179.171828][ T3731] ? kthread_blkcg+0x110/0x110 [ 179.172587][ T3731] ret_from_fork+0x1f/0x30 [ 179.173348][ T3731] [ 179.173752][ T3731] Uninit was created at: [ 179.174409][ T3731] kmsan_internal_poison_shadow+0x5c/0xf0 [ 179.175373][ T3731] kmsan_slab_free+0x76/0xc0 [ 179.176060][ T3731] kfree+0x3a5/0x1180 [ 179.176664][ T3731] __sk_destruct+0x8af/0xb80 [ 179.177375][ T3731] __sk_free+0x812/0x8c0 [ 179.178032][ T3731] sk_free+0x97/0x130 [ 179.178686][ T3731] l2cap_sock_release+0x3d5/0x4d0 [ 179.179457][ T3731] sock_close+0x150/0x450 [ 179.180117][ T3731] __fput+0x6bd/0xf00 [ 179.180787][ T3731] ____fput+0x37/0x40 [ 179.181481][ T3731] task_work_run+0x140/0x280 [ 179.182219][ T3731] do_exit+0xe51/0x3e60 [ 179.182930][ T3731] do_group_exit+0x20e/0x450 [ 179.183656][ T3731] get_signal+0x2dfb/0x38f0 [ 179.184344][ T3731] arch_do_signal_or_restart+0xaa/0xe10 [ 179.185266][ T3731] exit_to_user_mode_prepare+0x2d2/0x560 [ 179.186136][ T3731] syscall_exit_to_user_mode+0x35/0x60 [ 179.186984][ T3731] do_syscall_64+0xc5/0x140 [ 179.187681][ T3731] entry_SYSCALL_64_after_hwframe+0x44/0xae [ 179.188604][ T3731] ===================================================== In our case, there are two Thread A and B: Context: Thread A: Context: Thread B: l2cap_chan_timeout() __se_sys_shutdown() l2cap_chan_close() l2cap_sock_shutdown() l2cap_chan_del() l2cap_chan_close() l2cap_sock_teardown_cb() l2cap_sock_teardown_cb() Once l2cap_sock_teardown_cb() excuted, this sock will be marked as SOCK_ZAPPED, and can be treated as killable in l2cap_sock_kill() if sock_orphan() has excuted, at this time we close sock through sock_close() which end to call l2cap_sock_kill() like Thread C: Context: Thread C: sock_close() l2cap_sock_release() sock_orphan() l2cap_sock_kill() #free sock if refcnt is 1 If C completed, Once A or B reaches l2cap_sock_teardown_cb() again, use-after-free happened. We should set chan->data to NULL if sock is destructed, for telling teardown operation is not allowed in l2cap_sock_teardown_cb(), and also we should avoid killing an already killed socket in l2cap_sock_close_cb(). Signed-off-by: Wang ShaoBo <[email protected]> Signed-off-by: Luiz Augusto von Dentz <[email protected]> Signed-off-by: Marcel Holtmann <[email protected]>

correct rtl_op_set_tim and rtl_update_beacon implementation missing

e36433f

patch found at 'http://www.spinics.net/lists/linux-wireless/msg96128.html'

aejsmith pushed a commit to aejsmith/linux that referenced this pull request Jul 24, 2015

Merge pull request torvalds#54 from HarveyHunt/ci20-v3.18-mmc-regress…

17b6872

…ion-cgu Ci20 v3.18 MSC0 Performance Fix

logic10492 pushed a commit to logic10492/linux-amd-zen2 that referenced this pull request Jan 18, 2024

Merge pull request torvalds#54 from sched-ext/bpf-next-merge

ff56aee

Bpf next merge

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

correct rtl_op_set_tim and rtl_update_beacon implementation missing #54

correct rtl_op_set_tim and rtl_update_beacon implementation missing #54

jpbruyere commented Oct 31, 2013

correct rtl_op_set_tim and rtl_update_beacon implementation missing #54

Are you sure you want to change the base?

correct rtl_op_set_tim and rtl_update_beacon implementation missing #54

Conversation

jpbruyere commented Oct 31, 2013