From 5f074f3e192f10c9fade898b9b3b8812e3d83342 Mon Sep 17 00:00:00 2001 From: Nick Desaulniers Date: Fri, 5 Apr 2019 18:38:45 -0700 Subject: lib/string.c: implement a basic bcmp A recent optimization in Clang (r355672) lowers comparisons of the return value of memcmp against zero to comparisons of the return value of bcmp against zero. This helps some platforms that implement bcmp more efficiently than memcmp. glibc simply aliases bcmp to memcmp, but an optimized implementation is in the works. This results in linkage failures for all targets with Clang due to the undefined symbol. For now, just implement bcmp as a tailcail to memcmp to unbreak the build. This routine can be further optimized in the future. Other ideas discussed: * A weak alias was discussed, but breaks for architectures that define their own implementations of memcmp since aliases to declarations are not permitted (only definitions). Arch-specific memcmp implementations typically declare memcmp in C headers, but implement them in assembly. * -ffreestanding also is used sporadically throughout the kernel. * -fno-builtin-bcmp doesn't work when doing LTO. Link: https://bugs.llvm.org/show_bug.cgi?id=41035 Link: https://code.woboq.org/userspace/glibc/string/memcmp.c.html#bcmp Link: https://github.com/llvm/llvm-project/commit/8e16d73346f8091461319a7dfc4ddd18eedcff13 Link: https://github.com/ClangBuiltLinux/linux/issues/416 Link: http://lkml.kernel.org/r/20190313211335.165605-1-ndesaulniers@google.com Signed-off-by: Nick Desaulniers Reported-by: Nathan Chancellor Reported-by: Adhemerval Zanella Suggested-by: Arnd Bergmann Suggested-by: James Y Knight Suggested-by: Masahiro Yamada Suggested-by: Nathan Chancellor Suggested-by: Rasmus Villemoes Acked-by: Steven Rostedt (VMware) Reviewed-by: Nathan Chancellor Tested-by: Nathan Chancellor Reviewed-by: Masahiro Yamada Reviewed-by: Andy Shevchenko Cc: David Laight Cc: Rasmus Villemoes Cc: Namhyung Kim Cc: Greg Kroah-Hartman Cc: Alexander Shishkin Cc: Dan Williams Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/string.h | 3 +++ 1 file changed, 3 insertions(+) (limited to 'include/linux') diff --git a/include/linux/string.h b/include/linux/string.h index 7927b875f80c..6ab0a6fa512e 100644 --- a/include/linux/string.h +++ b/include/linux/string.h @@ -150,6 +150,9 @@ extern void * memscan(void *,int,__kernel_size_t); #ifndef __HAVE_ARCH_MEMCMP extern int memcmp(const void *,const void *,__kernel_size_t); #endif +#ifndef __HAVE_ARCH_BCMP +extern int bcmp(const void *,const void *,__kernel_size_t); +#endif #ifndef __HAVE_ARCH_MEMCHR extern void * memchr(const void *,int,__kernel_size_t); #endif -- cgit v1.2.3 From 6147e136ff5071609b54f18982dea87706288e21 Mon Sep 17 00:00:00 2001 From: Arnd Bergmann Date: Fri, 5 Apr 2019 18:38:53 -0700 Subject: include/linux/bitrev.h: fix constant bitrev clang points out with hundreds of warnings that the bitrev macros have a problem with constant input: drivers/hwmon/sht15.c:187:11: error: variable '__x' is uninitialized when used within its own initialization [-Werror,-Wuninitialized] u8 crc = bitrev8(data->val_status & 0x0F); ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ include/linux/bitrev.h:102:21: note: expanded from macro 'bitrev8' __constant_bitrev8(__x) : \ ~~~~~~~~~~~~~~~~~~~^~~~ include/linux/bitrev.h:67:11: note: expanded from macro '__constant_bitrev8' u8 __x = x; \ ~~~ ^ Both the bitrev and the __constant_bitrev macros use an internal variable named __x, which goes horribly wrong when passing one to the other. The obvious fix is to rename one of the variables, so this adds an extra '_'. It seems we got away with this because - there are only a few drivers using bitrev macros - usually there are no constant arguments to those - when they are constant, they tend to be either 0 or (unsigned)-1 (drivers/isdn/i4l/isdnhdlc.o, drivers/iio/amplifiers/ad8366.c) and give the correct result by pure chance. In fact, the only driver that I could find that gets different results with this is drivers/net/wan/slic_ds26522.c, which in turn is a driver for fairly rare hardware (adding the maintainer to Cc for testing). Link: http://lkml.kernel.org/r/20190322140503.123580-1-arnd@arndb.de Fixes: 556d2f055bf6 ("ARM: 8187/1: add CONFIG_HAVE_ARCH_BITREVERSE to support rbit instruction") Signed-off-by: Arnd Bergmann Reviewed-by: Nick Desaulniers Cc: Zhao Qiang Cc: Yalin Wang Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/bitrev.h | 46 +++++++++++++++++++++++----------------------- 1 file changed, 23 insertions(+), 23 deletions(-) (limited to 'include/linux') diff --git a/include/linux/bitrev.h b/include/linux/bitrev.h index 50fb0dee23e8..d35b8ec1c485 100644 --- a/include/linux/bitrev.h +++ b/include/linux/bitrev.h @@ -34,41 +34,41 @@ static inline u32 __bitrev32(u32 x) #define __constant_bitrev32(x) \ ({ \ - u32 __x = x; \ - __x = (__x >> 16) | (__x << 16); \ - __x = ((__x & (u32)0xFF00FF00UL) >> 8) | ((__x & (u32)0x00FF00FFUL) << 8); \ - __x = ((__x & (u32)0xF0F0F0F0UL) >> 4) | ((__x & (u32)0x0F0F0F0FUL) << 4); \ - __x = ((__x & (u32)0xCCCCCCCCUL) >> 2) | ((__x & (u32)0x33333333UL) << 2); \ - __x = ((__x & (u32)0xAAAAAAAAUL) >> 1) | ((__x & (u32)0x55555555UL) << 1); \ - __x; \ + u32 ___x = x; \ + ___x = (___x >> 16) | (___x << 16); \ + ___x = ((___x & (u32)0xFF00FF00UL) >> 8) | ((___x & (u32)0x00FF00FFUL) << 8); \ + ___x = ((___x & (u32)0xF0F0F0F0UL) >> 4) | ((___x & (u32)0x0F0F0F0FUL) << 4); \ + ___x = ((___x & (u32)0xCCCCCCCCUL) >> 2) | ((___x & (u32)0x33333333UL) << 2); \ + ___x = ((___x & (u32)0xAAAAAAAAUL) >> 1) | ((___x & (u32)0x55555555UL) << 1); \ + ___x; \ }) #define __constant_bitrev16(x) \ ({ \ - u16 __x = x; \ - __x = (__x >> 8) | (__x << 8); \ - __x = ((__x & (u16)0xF0F0U) >> 4) | ((__x & (u16)0x0F0FU) << 4); \ - __x = ((__x & (u16)0xCCCCU) >> 2) | ((__x & (u16)0x3333U) << 2); \ - __x = ((__x & (u16)0xAAAAU) >> 1) | ((__x & (u16)0x5555U) << 1); \ - __x; \ + u16 ___x = x; \ + ___x = (___x >> 8) | (___x << 8); \ + ___x = ((___x & (u16)0xF0F0U) >> 4) | ((___x & (u16)0x0F0FU) << 4); \ + ___x = ((___x & (u16)0xCCCCU) >> 2) | ((___x & (u16)0x3333U) << 2); \ + ___x = ((___x & (u16)0xAAAAU) >> 1) | ((___x & (u16)0x5555U) << 1); \ + ___x; \ }) #define __constant_bitrev8x4(x) \ ({ \ - u32 __x = x; \ - __x = ((__x & (u32)0xF0F0F0F0UL) >> 4) | ((__x & (u32)0x0F0F0F0FUL) << 4); \ - __x = ((__x & (u32)0xCCCCCCCCUL) >> 2) | ((__x & (u32)0x33333333UL) << 2); \ - __x = ((__x & (u32)0xAAAAAAAAUL) >> 1) | ((__x & (u32)0x55555555UL) << 1); \ - __x; \ + u32 ___x = x; \ + ___x = ((___x & (u32)0xF0F0F0F0UL) >> 4) | ((___x & (u32)0x0F0F0F0FUL) << 4); \ + ___x = ((___x & (u32)0xCCCCCCCCUL) >> 2) | ((___x & (u32)0x33333333UL) << 2); \ + ___x = ((___x & (u32)0xAAAAAAAAUL) >> 1) | ((___x & (u32)0x55555555UL) << 1); \ + ___x; \ }) #define __constant_bitrev8(x) \ ({ \ - u8 __x = x; \ - __x = (__x >> 4) | (__x << 4); \ - __x = ((__x & (u8)0xCCU) >> 2) | ((__x & (u8)0x33U) << 2); \ - __x = ((__x & (u8)0xAAU) >> 1) | ((__x & (u8)0x55U) << 1); \ - __x; \ + u8 ___x = x; \ + ___x = (___x >> 4) | (___x << 4); \ + ___x = ((___x & (u8)0xCCU) >> 2) | ((___x & (u8)0x33U) << 2); \ + ___x = ((___x & (u8)0xAAU) >> 1) | ((___x & (u8)0x55U) << 1); \ + ___x; \ }) #define bitrev32(x) \ -- cgit v1.2.3 From fcae96ff96538f66e7acd5d4e0f2e7516ff8cbd0 Mon Sep 17 00:00:00 2001 From: Jann Horn Date: Fri, 5 Apr 2019 18:39:01 -0700 Subject: mm: fix vm_fault_t cast in VM_FAULT_GET_HINDEX() Symmetrically to VM_FAULT_SET_HINDEX(), we need a force-cast in VM_FAULT_GET_HINDEX() to tell sparse that this is intentional. Sparse complains about the current code when building a kernel with CONFIG_MEMORY_FAILURE: arch/x86/mm/fault.c:1058:53: warning: restricted vm_fault_t degrades to integer Link: http://lkml.kernel.org/r/20190327204117.35215-1-jannh@google.com Fixes: 3d3539018d2c ("mm: create the new vm_fault_t type") Signed-off-by: Jann Horn Reviewed-by: Andrew Morton Cc: Souptick Joarder Cc: Matthew Wilcox Cc: Vlastimil Babka Cc: "Kirill A. Shutemov" Cc: Rik van Riel Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/mm_types.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h index 7eade9132f02..4ef4bbe78a1d 100644 --- a/include/linux/mm_types.h +++ b/include/linux/mm_types.h @@ -671,7 +671,7 @@ enum vm_fault_reason { /* Encode hstate index for a hwpoisoned large page */ #define VM_FAULT_SET_HINDEX(x) ((__force vm_fault_t)((x) << 16)) -#define VM_FAULT_GET_HINDEX(x) (((x) >> 16) & 0xf) +#define VM_FAULT_GET_HINDEX(x) (((__force unsigned int)(x) >> 16) & 0xf) #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | \ VM_FAULT_SIGSEGV | VM_FAULT_HWPOISON | \ -- cgit v1.2.3 From 0b3d6e6f2dd0a7b697b1aa8c167265908940624b Mon Sep 17 00:00:00 2001 From: Greg Thelen Date: Fri, 5 Apr 2019 18:39:18 -0700 Subject: mm: writeback: use exact memcg dirty counts Since commit a983b5ebee57 ("mm: memcontrol: fix excessive complexity in memory.stat reporting") memcg dirty and writeback counters are managed as: 1) per-memcg per-cpu values in range of [-32..32] 2) per-memcg atomic counter When a per-cpu counter cannot fit in [-32..32] it's flushed to the atomic. Stat readers only check the atomic. Thus readers such as balance_dirty_pages() may see a nontrivial error margin: 32 pages per cpu. Assuming 100 cpus: 4k x86 page_size: 13 MiB error per memcg 64k ppc page_size: 200 MiB error per memcg Considering that dirty+writeback are used together for some decisions the errors double. This inaccuracy can lead to undeserved oom kills. One nasty case is when all per-cpu counters hold positive values offsetting an atomic negative value (i.e. per_cpu[*]=32, atomic=n_cpu*-32). balance_dirty_pages() only consults the atomic and does not consider throttling the next n_cpu*32 dirty pages. If the file_lru is in the 13..200 MiB range then there's absolutely no dirty throttling, which burdens vmscan with only dirty+writeback pages thus resorting to oom kill. It could be argued that tiny containers are not supported, but it's more subtle. It's the amount the space available for file lru that matters. If a container has memory.max-200MiB of non reclaimable memory, then it will also suffer such oom kills on a 100 cpu machine. The following test reliably ooms without this patch. This patch avoids oom kills. $ cat test mount -t cgroup2 none /dev/cgroup cd /dev/cgroup echo +io +memory > cgroup.subtree_control mkdir test cd test echo 10M > memory.max (echo $BASHPID > cgroup.procs && exec /memcg-writeback-stress /foo) (echo $BASHPID > cgroup.procs && exec dd if=/dev/zero of=/foo bs=2M count=100) $ cat memcg-writeback-stress.c /* * Dirty pages from all but one cpu. * Clean pages from the non dirtying cpu. * This is to stress per cpu counter imbalance. * On a 100 cpu machine: * - per memcg per cpu dirty count is 32 pages for each of 99 cpus * - per memcg atomic is -99*32 pages * - thus the complete dirty limit: sum of all counters 0 * - balance_dirty_pages() only sees atomic count -99*32 pages, which * it max()s to 0. * - So a workload can dirty -99*32 pages before balance_dirty_pages() * cares. */ #define _GNU_SOURCE #include #include #include #include #include #include #include #include #include static char *buf; static int bufSize; static void set_affinity(int cpu) { cpu_set_t affinity; CPU_ZERO(&affinity); CPU_SET(cpu, &affinity); if (sched_setaffinity(0, sizeof(affinity), &affinity)) err(1, "sched_setaffinity"); } static void dirty_on(int output_fd, int cpu) { int i, wrote; set_affinity(cpu); for (i = 0; i < 32; i++) { for (wrote = 0; wrote < bufSize; ) { int ret = write(output_fd, buf+wrote, bufSize-wrote); if (ret == -1) err(1, "write"); wrote += ret; } } } int main(int argc, char **argv) { int cpu, flush_cpu = 1, output_fd; const char *output; if (argc != 2) errx(1, "usage: output_file"); output = argv[1]; bufSize = getpagesize(); buf = malloc(getpagesize()); if (buf == NULL) errx(1, "malloc failed"); output_fd = open(output, O_CREAT|O_RDWR); if (output_fd == -1) err(1, "open(%s)", output); for (cpu = 0; cpu < get_nprocs(); cpu++) { if (cpu != flush_cpu) dirty_on(output_fd, cpu); } set_affinity(flush_cpu); if (fsync(output_fd)) err(1, "fsync(%s)", output); if (close(output_fd)) err(1, "close(%s)", output); free(buf); } Make balance_dirty_pages() and wb_over_bg_thresh() work harder to collect exact per memcg counters. This avoids the aforementioned oom kills. This does not affect the overhead of memory.stat, which still reads the single atomic counter. Why not use percpu_counter? memcg already handles cpus going offline, so no need for that overhead from percpu_counter. And the percpu_counter spinlocks are more heavyweight than is required. It probably also makes sense to use exact dirty and writeback counters in memcg oom reports. But that is saved for later. Link: http://lkml.kernel.org/r/20190329174609.164344-1-gthelen@google.com Signed-off-by: Greg Thelen Reviewed-by: Roman Gushchin Acked-by: Johannes Weiner Cc: Michal Hocko Cc: Vladimir Davydov Cc: Tejun Heo Cc: [4.16+] Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- include/linux/memcontrol.h | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/memcontrol.h b/include/linux/memcontrol.h index 1f3d880b7ca1..dbb6118370c1 100644 --- a/include/linux/memcontrol.h +++ b/include/linux/memcontrol.h @@ -566,7 +566,10 @@ struct mem_cgroup *lock_page_memcg(struct page *page); void __unlock_page_memcg(struct mem_cgroup *memcg); void unlock_page_memcg(struct page *page); -/* idx can be of type enum memcg_stat_item or node_stat_item */ +/* + * idx can be of type enum memcg_stat_item or node_stat_item. + * Keep in sync with memcg_exact_page_state(). + */ static inline unsigned long memcg_page_state(struct mem_cgroup *memcg, int idx) { -- cgit v1.2.3