Searched +hist:2 +hist:c2345c2 (Results 1 - 2 of 2) sorted by relevance
| /linux-master/include/linux/ | ||
| H A D | types.h | diff 2e346b19 Mon Dec 11 12:11:05 MST 2023 Kent Overstreet <kent.overstreet@linux.dev> ktime.h: move ktime_t to types.h ktime.h pulls in quite a few headers recursively (including printk.h) - this is going to help with trimming sched.h dependencies. Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev> diff ee1ee6db Thu Mar 23 14:55:31 MDT 2023 Thomas Gleixner <tglx@linutronix.de> atomics: Provide rcuref - scalable reference counting atomic_t based reference counting, including refcount_t, uses atomic_inc_not_zero() for acquiring a reference. atomic_inc_not_zero() is implemented with a atomic_try_cmpxchg() loop. High contention of the reference count leads to retry loops and scales badly. There is nothing to improve on this implementation as the semantics have to be preserved. Provide rcuref as a scalable alternative solution which is suitable for RCU managed objects. Similar to refcount_t it comes with overflow and underflow detection and mitigation. rcuref treats the underlying atomic_t as an unsigned integer and partitions this space into zones: 0x00000000 - 0x7FFFFFFF valid zone (1 .. (INT_MAX + 1) references) 0x80000000 - 0xBFFFFFFF saturation zone 0xC0000000 - 0xFFFFFFFE dead zone 0xFFFFFFFF no reference rcuref_get() unconditionally increments the reference count with atomic_add_negative_relaxed(). rcuref_put() unconditionally decrements the reference count with atomic_add_negative_release(). This unconditional increment avoids the inc_not_zero() problem, but requires a more complex implementation on the put() side when the count drops from 0 to -1. When this transition is detected then it is attempted to mark the reference count dead, by setting it to the midpoint of the dead zone with a single atomic_cmpxchg_release() operation. This operation can fail due to a concurrent rcuref_get() elevating the reference count from -1 to 0 again. If the unconditional increment in rcuref_get() hits a reference count which is marked dead (or saturated) it will detect it after the fact and bring back the reference count to the midpoint of the respective zone. The zones provide enough tolerance which makes it practically impossible to escape from a zone. The racy implementation of rcuref_put() requires to protect rcuref_put() against a grace period ending in order to prevent a subtle use after free. As RCU is the only mechanism which allows to protect against that, it is not possible to fully replace the atomic_inc_not_zero() based implementation of refcount_t with this scheme. The final drop is slightly more expensive than the atomic_dec_return() counterpart, but that's not the case which this is optimized for. The optimization is on the high frequeunt get()/put() pairs and their scalability. The performance of an uncontended rcuref_get()/put() pair where the put() is not dropping the last reference is still on par with the plain atomic operations, while at the same time providing overflow and underflow detection and mitigation. The performance of rcuref compared to plain atomic_inc_not_zero() and atomic_dec_return() based reference counting under contention: - Micro benchmark: All CPUs running a increment/decrement loop on an elevated reference count, which means the 0 to -1 transition never happens. The performance gain depends on microarchitecture and the number of CPUs and has been observed in the range of 1.3X to 4.7X - Conversion of dst_entry::__refcnt to rcuref and testing with the localhost memtier/memcached benchmark. That benchmark shows the reference count contention prominently. The performance gain depends on microarchitecture and the number of CPUs and has been observed in the range of 1.1X to 2.6X over the previous fix for the false sharing issue vs. struct dst_entry::__refcnt. When memtier is run over a real 1Gb network connection, there is a small gain on top of the false sharing fix. The two changes combined result in a 2%-5% total gain for that networked test. Reported-by: Wangyang Guo <wangyang.guo@intel.com> Reported-by: Arjan Van De Ven <arjan.van.de.ven@intel.com> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Link: https://lore.kernel.org/r/20230323102800.158429195@linutronix.de diff a0019cd7 Wed Mar 16 06:24:07 MDT 2022 Jiri Olsa <jolsa@kernel.org> lib/sort: Add priv pointer to swap function Adding support to have priv pointer in swap callback function. Following the initial change on cmp callback functions [1] and adding SWAP_WRAPPER macro to identify sort call of sort_r. Signed-off-by: Jiri Olsa <jolsa@kernel.org> Signed-off-by: Alexei Starovoitov <ast@kernel.org> Reviewed-by: Masami Hiramatsu <mhiramat@kernel.org> Link: https://lore.kernel.org/bpf/20220316122419.933957-2-jolsa@kernel.org [1] 4333fb96ca10 ("media: lib/sort.c: implement sort() variant taking context argument") diff 2a785996 Tue Nov 05 03:10:01 MST 2019 Arnd Bergmann <arnd@arndb.de> y2038: uapi: change __kernel_time_t to __kernel_old_time_t This is mainly a patch for clarification, and to let us remove the time_t definition from the kernel to prevent new users from creeping in that might not be y2038-safe. All remaining uses of 'time_t' or '__kernel_time_t' are part of the user API that cannot be changed by that either have a replacement or that do not suffer from the y2038 overflow. Acked-by: Deepa Dinamani <deepa.kernel@gmail.com> Acked-by: Christian Brauner <christian.brauner@ubuntu.com> Signed-off-by: Arnd Bergmann <arnd@arndb.de> diff 72deb455 Fri Apr 05 10:08:59 MDT 2019 Christoph Hellwig <hch@lst.de> block: remove CONFIG_LBDAF Currently support for 64-bit sector_t and blkcnt_t is optional on 32-bit architectures. These types are required to support block device and/or file sizes larger than 2 TiB, and have generally defaulted to on for a long time. Enabling the option only increases the i386 tinyconfig size by 145 bytes, and many data structures already always use 64-bit values for their in-core and on-disk data structures anyway, so there should not be a large change in dynamic memory usage either. Dropping this option removes a somewhat weird non-default config that has cause various bugs or compiler warnings when actually used. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jens Axboe <axboe@kernel.dk> diff 6d0e8d53 Thu Jun 07 18:10:27 MDT 2018 Masahiro Yamada <yamada.masahiro@socionext.com> include/linux/types.h: define aligned_ types based on uapi header <uapi/linux/types.h> has the same typedefs except that it prefixes them with double-underscore for user space. Use them for the kernel space typedefs. Link: http://lkml.kernel.org/r/1526350925-14922-2-git-send-email-yamada.masahiro@socionext.com Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com> Reviewed-by: Andrew Morton <akpm@linux-foundation.org> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Alexey Dobriyan <adobriyan@gmail.com> Cc: Lihao Liang <lianglihao@huawei.com> Cc: Philippe Ombredanne <pombredanne@nexb.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> diff b2441318 Wed Nov 01 08:07:57 MDT 2017 Greg Kroah-Hartman <gregkh@linuxfoundation.org> License cleanup: add SPDX GPL-2.0 license identifier to files with no license Many source files in the tree are missing licensing information, which makes it harder for compliance tools to determine the correct license. By default all files without license information are under the default license of the kernel, which is GPL version 2. Update the files which contain no license information with the 'GPL-2.0' SPDX license identifier. The SPDX identifier is a legally binding shorthand, which can be used instead of the full boiler plate text. This patch is based on work done by Thomas Gleixner and Kate Stewart and Philippe Ombredanne. How this work was done: Patches were generated and checked against linux-4.14-rc6 for a subset of the use cases: - file had no licensing information it it. - file was a */uapi/* one with no licensing information in it, - file was a */uapi/* one with existing licensing information, Further patches will be generated in subsequent months to fix up cases where non-standard license headers were used, and references to license had to be inferred by heuristics based on keywords. The analysis to determine which SPDX License Identifier to be applied to a file was done in a spreadsheet of side by side results from of the output of two independent scanners (ScanCode & Windriver) producing SPDX tag:value files created by Philippe Ombredanne. Philippe prepared the base worksheet, and did an initial spot review of a few 1000 files. The 4.13 kernel was the starting point of the analysis with 60,537 files assessed. Kate Stewart did a file by file comparison of the scanner results in the spreadsheet to determine which SPDX license identifier(s) to be applied to the file. She confirmed any determination that was not immediately clear with lawyers working with the Linux Foundation. Criteria used to select files for SPDX license identifier tagging was: - Files considered eligible had to be source code files. - Make and config files were included as candidates if they contained >5 lines of source - File already had some variant of a license header in it (even if <5 lines). All documentation files were explicitly excluded. The following heuristics were used to determine which SPDX license identifiers to apply. - when both scanners couldn't find any license traces, file was considered to have no license information in it, and the top level COPYING file license applied. For non */uapi/* files that summary was: SPDX license identifier # files ---------------------------------------------------|------- GPL-2.0 11139 and resulted in the first patch in this series. If that file was a */uapi/* path one, it was "GPL-2.0 WITH Linux-syscall-note" otherwise it was "GPL-2.0". Results of that was: SPDX license identifier # files ---------------------------------------------------|------- GPL-2.0 WITH Linux-syscall-note 930 and resulted in the second patch in this series. - if a file had some form of licensing information in it, and was one of the */uapi/* ones, it was denoted with the Linux-syscall-note if any GPL family license was found in the file or had no licensing in it (per prior point). Results summary: SPDX license identifier # files ---------------------------------------------------|------ GPL-2.0 WITH Linux-syscall-note 270 GPL-2.0+ WITH Linux-syscall-note 169 ((GPL-2.0 WITH Linux-syscall-note) OR BSD-2-Clause) 21 ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) 17 LGPL-2.1+ WITH Linux-syscall-note 15 GPL-1.0+ WITH Linux-syscall-note 14 ((GPL-2.0+ WITH Linux-syscall-note) OR BSD-3-Clause) 5 LGPL-2.0+ WITH Linux-syscall-note 4 LGPL-2.1 WITH Linux-syscall-note 3 ((GPL-2.0 WITH Linux-syscall-note) OR MIT) 3 ((GPL-2.0 WITH Linux-syscall-note) AND MIT) 1 and that resulted in the third patch in this series. - when the two scanners agreed on the detected license(s), that became the concluded license(s). - when there was disagreement between the two scanners (one detected a license but the other didn't, or they both detected different licenses) a manual inspection of the file occurred. - In most cases a manual inspection of the information in the file resulted in a clear resolution of the license that should apply (and which scanner probably needed to revisit its heuristics). - When it was not immediately clear, the license identifier was confirmed with lawyers working with the Linux Foundation. - If there was any question as to the appropriate license identifier, the file was flagged for further research and to be revisited later in time. In total, over 70 hours of logged manual review was done on the spreadsheet to determine the SPDX license identifiers to apply to the source files by Kate, Philippe, Thomas and, in some cases, confirmation by lawyers working with the Linux Foundation. Kate also obtained a third independent scan of the 4.13 code base from FOSSology, and compared selected files where the other two scanners disagreed against that SPDX file, to see if there was new insights. The Windriver scanner is based on an older version of FOSSology in part, so they are related. Thomas did random spot checks in about 500 files from the spreadsheets for the uapi headers and agreed with SPDX license identifier in the files he inspected. For the non-uapi files Thomas did random spot checks in about 15000 files. In initial set of patches against 4.14-rc6, 3 files were found to have copy/paste license identifier errors, and have been fixed to reflect the correct identifier. Additionally Philippe spent 10 hours this week doing a detailed manual inspection and review of the 12,461 patched files from the initial patch version early this week with: - a full scancode scan run, collecting the matched texts, detected license ids and scores - reviewing anything where there was a license detected (about 500+ files) to ensure that the applied SPDX license was correct - reviewing anything where there was no detection but the patch license was not GPL-2.0 WITH Linux-syscall-note to ensure that the applied SPDX license was correct This produced a worksheet with 20 files needing minor correction. This worksheet was then exported into 3 different .csv files for the different types of files to be modified. These .csv files were then reviewed by Greg. Thomas wrote a script to parse the csv files and add the proper SPDX tag to the file, in the format that the file expected. This script was further refined by Greg based on the output to detect more types of files automatically and to distinguish between header and source .c files (which need different comment types.) Finally Greg ran the script using the .csv files to generate the patches. Reviewed-by: Kate Stewart <kstewart@linuxfoundation.org> Reviewed-by: Philippe Ombredanne <pombredanne@nexb.com> Reviewed-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> diff b2441318 Wed Nov 01 08:07:57 MDT 2017 Greg Kroah-Hartman <gregkh@linuxfoundation.org> License cleanup: add SPDX GPL-2.0 license identifier to files with no license Many source files in the tree are missing licensing information, which makes it harder for compliance tools to determine the correct license. By default all files without license information are under the default license of the kernel, which is GPL version 2. Update the files which contain no license information with the 'GPL-2.0' SPDX license identifier. The SPDX identifier is a legally binding shorthand, which can be used instead of the full boiler plate text. This patch is based on work done by Thomas Gleixner and Kate Stewart and Philippe Ombredanne. How this work was done: Patches were generated and checked against linux-4.14-rc6 for a subset of the use cases: - file had no licensing information it it. - file was a */uapi/* one with no licensing information in it, - file was a */uapi/* one with existing licensing information, Further patches will be generated in subsequent months to fix up cases where non-standard license headers were used, and references to license had to be inferred by heuristics based on keywords. The analysis to determine which SPDX License Identifier to be applied to a file was done in a spreadsheet of side by side results from of the output of two independent scanners (ScanCode & Windriver) producing SPDX tag:value files created by Philippe Ombredanne. Philippe prepared the base worksheet, and did an initial spot review of a few 1000 files. The 4.13 kernel was the starting point of the analysis with 60,537 files assessed. Kate Stewart did a file by file comparison of the scanner results in the spreadsheet to determine which SPDX license identifier(s) to be applied to the file. She confirmed any determination that was not immediately clear with lawyers working with the Linux Foundation. Criteria used to select files for SPDX license identifier tagging was: - Files considered eligible had to be source code files. - Make and config files were included as candidates if they contained >5 lines of source - File already had some variant of a license header in it (even if <5 lines). All documentation files were explicitly excluded. The following heuristics were used to determine which SPDX license identifiers to apply. - when both scanners couldn't find any license traces, file was considered to have no license information in it, and the top level COPYING file license applied. For non */uapi/* files that summary was: SPDX license identifier # files ---------------------------------------------------|------- GPL-2.0 11139 and resulted in the first patch in this series. If that file was a */uapi/* path one, it was "GPL-2.0 WITH Linux-syscall-note" otherwise it was "GPL-2.0". Results of that was: SPDX license identifier # files ---------------------------------------------------|------- GPL-2.0 WITH Linux-syscall-note 930 and resulted in the second patch in this series. - if a file had some form of licensing information in it, and was one of the */uapi/* ones, it was denoted with the Linux-syscall-note if any GPL family license was found in the file or had no licensing in it (per prior point). Results summary: SPDX license identifier # files ---------------------------------------------------|------ GPL-2.0 WITH Linux-syscall-note 270 GPL-2.0+ WITH Linux-syscall-note 169 ((GPL-2.0 WITH Linux-syscall-note) OR BSD-2-Clause) 21 ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) 17 LGPL-2.1+ WITH Linux-syscall-note 15 GPL-1.0+ WITH Linux-syscall-note 14 ((GPL-2.0+ WITH Linux-syscall-note) OR BSD-3-Clause) 5 LGPL-2.0+ WITH Linux-syscall-note 4 LGPL-2.1 WITH Linux-syscall-note 3 ((GPL-2.0 WITH Linux-syscall-note) OR MIT) 3 ((GPL-2.0 WITH Linux-syscall-note) AND MIT) 1 and that resulted in the third patch in this series. - when the two scanners agreed on the detected license(s), that became the concluded license(s). - when there was disagreement between the two scanners (one detected a license but the other didn't, or they both detected different licenses) a manual inspection of the file occurred. - In most cases a manual inspection of the information in the file resulted in a clear resolution of the license that should apply (and which scanner probably needed to revisit its heuristics). - When it was not immediately clear, the license identifier was confirmed with lawyers working with the Linux Foundation. - If there was any question as to the appropriate license identifier, the file was flagged for further research and to be revisited later in time. In total, over 70 hours of logged manual review was done on the spreadsheet to determine the SPDX license identifiers to apply to the source files by Kate, Philippe, Thomas and, in some cases, confirmation by lawyers working with the Linux Foundation. Kate also obtained a third independent scan of the 4.13 code base from FOSSology, and compared selected files where the other two scanners disagreed against that SPDX file, to see if there was new insights. The Windriver scanner is based on an older version of FOSSology in part, so they are related. Thomas did random spot checks in about 500 files from the spreadsheets for the uapi headers and agreed with SPDX license identifier in the files he inspected. For the non-uapi files Thomas did random spot checks in about 15000 files. In initial set of patches against 4.14-rc6, 3 files were found to have copy/paste license identifier errors, and have been fixed to reflect the correct identifier. Additionally Philippe spent 10 hours this week doing a detailed manual inspection and review of the 12,461 patched files from the initial patch version early this week with: - a full scancode scan run, collecting the matched texts, detected license ids and scores - reviewing anything where there was a license detected (about 500+ files) to ensure that the applied SPDX license was correct - reviewing anything where there was no detection but the patch license was not GPL-2.0 WITH Linux-syscall-note to ensure that the applied SPDX license was correct This produced a worksheet with 20 files needing minor correction. This worksheet was then exported into 3 different .csv files for the different types of files to be modified. These .csv files were then reviewed by Greg. Thomas wrote a script to parse the csv files and add the proper SPDX tag to the file, in the format that the file expected. This script was further refined by Greg based on the output to detect more types of files automatically and to distinguish between header and source .c files (which need different comment types.) Finally Greg ran the script using the .csv files to generate the patches. Reviewed-by: Kate Stewart <kstewart@linuxfoundation.org> Reviewed-by: Philippe Ombredanne <pombredanne@nexb.com> Reviewed-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> diff ec90a194 Wed Jun 10 01:53:06 MDT 2015 Paul E. McKenney <paulmck@kernel.org> rcu: Create a synchronize_rcu_mult() There have been several requests for a primitive that waits for grace periods for several RCU flavors concurrently, so this commit creates it. This is a variadic macro, and you pass in the call_rcu() functions of the flavors of RCU that you wish to wait for. Note that you cannot pass in call_srcu() for two reasons: (1) This would result in a type mismatch and (2) You need to specify which srcu_struct you want to use. Handle this by creating a wrapper function for your SRCU domain, for example: void call_srcu_mine(struct rcu_head *head, rcu_callback_t func) { call_srcu(&ss_mine, head, func); } You can then do something like this: synchronize_rcu_mult(call_srcu_mine, call_rcu, call_rcu_sched); Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> diff 81880d60 Sun May 16 22:34:57 MDT 2010 Anton Blanchard <anton@samba.org> atomic_t: Remove volatile from atomic_t definition When looking at a performance problem on PowerPC, I noticed some awful code generation: c00000000051fc98: 3b 60 00 01 li r27,1 ... c00000000051fca0: 3b 80 00 00 li r28,0 ... c00000000051fcdc: 93 61 00 70 stw r27,112(r1) c00000000051fce0: 93 81 00 74 stw r28,116(r1) c00000000051fce4: 81 21 00 70 lwz r9,112(r1) c00000000051fce8: 80 01 00 74 lwz r0,116(r1) c00000000051fcec: 7d 29 07 b4 extsw r9,r9 c00000000051fcf0: 7c 00 07 b4 extsw r0,r0 c00000000051fcf4: 7c 20 04 ac lwsync c00000000051fcf8: 7d 60 f8 28 lwarx r11,0,r31 c00000000051fcfc: 7c 0b 48 00 cmpw r11,r9 c00000000051fd00: 40 c2 00 10 bne- c00000000051fd10 c00000000051fd04: 7c 00 f9 2d stwcx. r0,0,r31 c00000000051fd08: 40 c2 ff f0 bne+ c00000000051fcf8 c00000000051fd0c: 4c 00 01 2c isync We create two constants, write them out to the stack, read them straight back in and sign extend them. What a mess. It turns out this bad code is a result of us defining atomic_t as a volatile int. We removed the volatile attribute from the powerpc atomic_t definition years ago, but commit ea435467500612636f8f4fb639ff6e76b2496e4b (atomic_t: unify all arch definitions) added it back in. To dig up an old quote from Linus: > The fact is, volatile on data structures is a bug. It's a wart in the C > language. It shouldn't be used. > > Volatile accesses in *code* can be ok, and if we have "atomic_read()" > expand to a "*(volatile int *)&(x)->value", then I'd be ok with that. > > But marking data structures volatile just makes the compiler screw up > totally, and makes code for initialization sequences etc much worse. And screw up it does :) With the volatile removed, we see much more reasonable code generation: c00000000051f5b8: 3b 60 00 01 li r27,1 ... c00000000051f5c0: 3b 80 00 00 li r28,0 ... c00000000051fc7c: 7c 20 04 ac lwsync c00000000051fc80: 7c 00 f8 28 lwarx r0,0,r31 c00000000051fc84: 7c 00 d8 00 cmpw r0,r27 c00000000051fc88: 40 c2 00 10 bne- c00000000051fc98 c00000000051fc8c: 7f 80 f9 2d stwcx. r28,0,r31 c00000000051fc90: 40 c2 ff f0 bne+ c00000000051fc80 c00000000051fc94: 4c 00 01 2c isync Six instructions less. Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> diff 81880d60 Sun May 16 22:34:57 MDT 2010 Anton Blanchard <anton@samba.org> atomic_t: Remove volatile from atomic_t definition When looking at a performance problem on PowerPC, I noticed some awful code generation: c00000000051fc98: 3b 60 00 01 li r27,1 ... c00000000051fca0: 3b 80 00 00 li r28,0 ... c00000000051fcdc: 93 61 00 70 stw r27,112(r1) c00000000051fce0: 93 81 00 74 stw r28,116(r1) c00000000051fce4: 81 21 00 70 lwz r9,112(r1) c00000000051fce8: 80 01 00 74 lwz r0,116(r1) c00000000051fcec: 7d 29 07 b4 extsw r9,r9 c00000000051fcf0: 7c 00 07 b4 extsw r0,r0 c00000000051fcf4: 7c 20 04 ac lwsync c00000000051fcf8: 7d 60 f8 28 lwarx r11,0,r31 c00000000051fcfc: 7c 0b 48 00 cmpw r11,r9 c00000000051fd00: 40 c2 00 10 bne- c00000000051fd10 c00000000051fd04: 7c 00 f9 2d stwcx. r0,0,r31 c00000000051fd08: 40 c2 ff f0 bne+ c00000000051fcf8 c00000000051fd0c: 4c 00 01 2c isync We create two constants, write them out to the stack, read them straight back in and sign extend them. What a mess. It turns out this bad code is a result of us defining atomic_t as a volatile int. We removed the volatile attribute from the powerpc atomic_t definition years ago, but commit ea435467500612636f8f4fb639ff6e76b2496e4b (atomic_t: unify all arch definitions) added it back in. To dig up an old quote from Linus: > The fact is, volatile on data structures is a bug. It's a wart in the C > language. It shouldn't be used. > > Volatile accesses in *code* can be ok, and if we have "atomic_read()" > expand to a "*(volatile int *)&(x)->value", then I'd be ok with that. > > But marking data structures volatile just makes the compiler screw up > totally, and makes code for initialization sequences etc much worse. And screw up it does :) With the volatile removed, we see much more reasonable code generation: c00000000051f5b8: 3b 60 00 01 li r27,1 ... c00000000051f5c0: 3b 80 00 00 li r28,0 ... c00000000051fc7c: 7c 20 04 ac lwsync c00000000051fc80: 7c 00 f8 28 lwarx r0,0,r31 c00000000051fc84: 7c 00 d8 00 cmpw r0,r27 c00000000051fc88: 40 c2 00 10 bne- c00000000051fc98 c00000000051fc8c: 7f 80 f9 2d stwcx. r28,0,r31 c00000000051fc90: 40 c2 ff f0 bne+ c00000000051fc80 c00000000051fc94: 4c 00 01 2c isync Six instructions less. Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> diff 81880d60 Sun May 16 22:34:57 MDT 2010 Anton Blanchard <anton@samba.org> atomic_t: Remove volatile from atomic_t definition When looking at a performance problem on PowerPC, I noticed some awful code generation: c00000000051fc98: 3b 60 00 01 li r27,1 ... c00000000051fca0: 3b 80 00 00 li r28,0 ... c00000000051fcdc: 93 61 00 70 stw r27,112(r1) c00000000051fce0: 93 81 00 74 stw r28,116(r1) c00000000051fce4: 81 21 00 70 lwz r9,112(r1) c00000000051fce8: 80 01 00 74 lwz r0,116(r1) c00000000051fcec: 7d 29 07 b4 extsw r9,r9 c00000000051fcf0: 7c 00 07 b4 extsw r0,r0 c00000000051fcf4: 7c 20 04 ac lwsync c00000000051fcf8: 7d 60 f8 28 lwarx r11,0,r31 c00000000051fcfc: 7c 0b 48 00 cmpw r11,r9 c00000000051fd00: 40 c2 00 10 bne- c00000000051fd10 c00000000051fd04: 7c 00 f9 2d stwcx. r0,0,r31 c00000000051fd08: 40 c2 ff f0 bne+ c00000000051fcf8 c00000000051fd0c: 4c 00 01 2c isync We create two constants, write them out to the stack, read them straight back in and sign extend them. What a mess. It turns out this bad code is a result of us defining atomic_t as a volatile int. We removed the volatile attribute from the powerpc atomic_t definition years ago, but commit ea435467500612636f8f4fb639ff6e76b2496e4b (atomic_t: unify all arch definitions) added it back in. To dig up an old quote from Linus: > The fact is, volatile on data structures is a bug. It's a wart in the C > language. It shouldn't be used. > > Volatile accesses in *code* can be ok, and if we have "atomic_read()" > expand to a "*(volatile int *)&(x)->value", then I'd be ok with that. > > But marking data structures volatile just makes the compiler screw up > totally, and makes code for initialization sequences etc much worse. And screw up it does :) With the volatile removed, we see much more reasonable code generation: c00000000051f5b8: 3b 60 00 01 li r27,1 ... c00000000051f5c0: 3b 80 00 00 li r28,0 ... c00000000051fc7c: 7c 20 04 ac lwsync c00000000051fc80: 7c 00 f8 28 lwarx r0,0,r31 c00000000051fc84: 7c 00 d8 00 cmpw r0,r27 c00000000051fc88: 40 c2 00 10 bne- c00000000051fc98 c00000000051fc8c: 7f 80 f9 2d stwcx. r28,0,r31 c00000000051fc90: 40 c2 ff f0 bne+ c00000000051fc80 c00000000051fc94: 4c 00 01 2c isync Six instructions less. Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> diff 81880d60 Sun May 16 22:34:57 MDT 2010 Anton Blanchard <anton@samba.org> atomic_t: Remove volatile from atomic_t definition When looking at a performance problem on PowerPC, I noticed some awful code generation: c00000000051fc98: 3b 60 00 01 li r27,1 ... c00000000051fca0: 3b 80 00 00 li r28,0 ... c00000000051fcdc: 93 61 00 70 stw r27,112(r1) c00000000051fce0: 93 81 00 74 stw r28,116(r1) c00000000051fce4: 81 21 00 70 lwz r9,112(r1) c00000000051fce8: 80 01 00 74 lwz r0,116(r1) c00000000051fcec: 7d 29 07 b4 extsw r9,r9 c00000000051fcf0: 7c 00 07 b4 extsw r0,r0 c00000000051fcf4: 7c 20 04 ac lwsync c00000000051fcf8: 7d 60 f8 28 lwarx r11,0,r31 c00000000051fcfc: 7c 0b 48 00 cmpw r11,r9 c00000000051fd00: 40 c2 00 10 bne- c00000000051fd10 c00000000051fd04: 7c 00 f9 2d stwcx. r0,0,r31 c00000000051fd08: 40 c2 ff f0 bne+ c00000000051fcf8 c00000000051fd0c: 4c 00 01 2c isync We create two constants, write them out to the stack, read them straight back in and sign extend them. What a mess. It turns out this bad code is a result of us defining atomic_t as a volatile int. We removed the volatile attribute from the powerpc atomic_t definition years ago, but commit ea435467500612636f8f4fb639ff6e76b2496e4b (atomic_t: unify all arch definitions) added it back in. To dig up an old quote from Linus: > The fact is, volatile on data structures is a bug. It's a wart in the C > language. It shouldn't be used. > > Volatile accesses in *code* can be ok, and if we have "atomic_read()" > expand to a "*(volatile int *)&(x)->value", then I'd be ok with that. > > But marking data structures volatile just makes the compiler screw up > totally, and makes code for initialization sequences etc much worse. And screw up it does :) With the volatile removed, we see much more reasonable code generation: c00000000051f5b8: 3b 60 00 01 li r27,1 ... c00000000051f5c0: 3b 80 00 00 li r28,0 ... c00000000051fc7c: 7c 20 04 ac lwsync c00000000051fc80: 7c 00 f8 28 lwarx r0,0,r31 c00000000051fc84: 7c 00 d8 00 cmpw r0,r27 c00000000051fc88: 40 c2 00 10 bne- c00000000051fc98 c00000000051fc8c: 7f 80 f9 2d stwcx. r28,0,r31 c00000000051fc90: 40 c2 ff f0 bne+ c00000000051fc80 c00000000051fc94: 4c 00 01 2c isync Six instructions less. Signed-off-by: Anton Blanchard <anton@samba.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> |
| H A D | bio.h | diff 492c5d45 Thu Nov 30 14:53:06 MST 2023 Keith Busch <kbusch@kernel.org> block: bio-integrity: directly map user buffers Passthrough commands that utilize metadata currently need to bounce the user space buffer through the kernel. Add support for mapping user space directly so that we can avoid this costly overhead. This is similar to how the normal bio data payload utilizes user addresses with bio_map_user_iov(). If the user address can't directly be used for reason, like too many segments or address unalignement, fallback to a copy of the user vec while keeping the user address pinned for the IO duration so that it can safely be copied on completion in any process context. Signed-off-by: Keith Busch <kbusch@kernel.org> Link: https://lore.kernel.org/r/20231130215309.2923568-2-kbusch@meta.com [axboe: fold in fix from Kanchan Joshi] Signed-off-by: Jens Axboe <axboe@kernel.dk> diff 2bc05769 Tue Aug 08 11:06:17 MDT 2023 Jens Axboe <axboe@kernel.dk> block: don't make REQ_POLLED imply REQ_NOWAIT Normally these two flags do go together, as the issuer of polled IO generally cannot wait for resources that will get freed as part of IO completion. This is because that very task is the one that will complete the request and free those resources, hence that would introduce a deadlock. But it is possible to have someone else issue the polled IO, eg via io_uring if the request is punted to io-wq. For that case, it's fine to have the task block on IO submission, as it is not the same task that will be completing the IO. It's completely up to the caller to ask for both polled and nowait IO separately! If we don't allow polled IO where IOCB_NOWAIT isn't set in the kiocb, then we can run into repeated -EAGAIN submissions and not make any progress. Reviewed-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Jens Axboe <axboe@kernel.dk> diff 09e8c253 Mon May 22 14:57:40 MDT 2023 David Howells <dhowells@redhat.com> block: Fix bio_flagged() so that gcc can better optimise it Fix bio_flagged() so that multiple instances of it, such as: if (bio_flagged(bio, BIO_PAGE_REFFED) || bio_flagged(bio, BIO_PAGE_PINNED)) can be combined by the gcc optimiser into a single test in assembly (arguably, this is a compiler optimisation issue[1]). The missed optimisation stems from bio_flagged() comparing the result of the bitwise-AND to zero. This results in an out-of-line bio_release_page() being compiled to something like: <+0>: mov 0x14(%rdi),%eax <+3>: test $0x1,%al <+5>: jne 0xffffffff816dac53 <bio_release_pages+11> <+7>: test $0x2,%al <+9>: je 0xffffffff816dac5c <bio_release_pages+20> <+11>: movzbl %sil,%esi <+15>: jmp 0xffffffff816daba1 <__bio_release_pages> <+20>: jmp 0xffffffff81d0b800 <__x86_return_thunk> However, the test is superfluous as the return type is bool. Removing it results in: <+0>: testb $0x3,0x14(%rdi) <+4>: je 0xffffffff816e4af4 <bio_release_pages+15> <+6>: movzbl %sil,%esi <+10>: jmp 0xffffffff816dab7c <__bio_release_pages> <+15>: jmp 0xffffffff81d0b7c0 <__x86_return_thunk> instead. Also, the MOVZBL instruction looks unnecessary[2] - I think it's just 're-booling' the mark_dirty parameter. Signed-off-by: David Howells <dhowells@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: John Hubbard <jhubbard@nvidia.com> cc: Jens Axboe <axboe@kernel.dk> cc: linux-block@vger.kernel.org Link: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108370 [1] Link: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108371 [2] Link: https://lore.kernel.org/r/167391056756.2311931.356007731815807265.stgit@warthog.procyon.org.uk/ # v6 Reviewed-by: Christian Brauner <brauner@kernel.org> Reviewed-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20230522205744.2825689-3-dhowells@redhat.com Signed-off-by: Jens Axboe <axboe@kernel.dk> diff 09e8c253 Mon May 22 14:57:40 MDT 2023 David Howells <dhowells@redhat.com> block: Fix bio_flagged() so that gcc can better optimise it Fix bio_flagged() so that multiple instances of it, such as: if (bio_flagged(bio, BIO_PAGE_REFFED) || bio_flagged(bio, BIO_PAGE_PINNED)) can be combined by the gcc optimiser into a single test in assembly (arguably, this is a compiler optimisation issue[1]). The missed optimisation stems from bio_flagged() comparing the result of the bitwise-AND to zero. This results in an out-of-line bio_release_page() being compiled to something like: <+0>: mov 0x14(%rdi),%eax <+3>: test $0x1,%al <+5>: jne 0xffffffff816dac53 <bio_release_pages+11> <+7>: test $0x2,%al <+9>: je 0xffffffff816dac5c <bio_release_pages+20> <+11>: movzbl %sil,%esi <+15>: jmp 0xffffffff816daba1 <__bio_release_pages> <+20>: jmp 0xffffffff81d0b800 <__x86_return_thunk> However, the test is superfluous as the return type is bool. Removing it results in: <+0>: testb $0x3,0x14(%rdi) <+4>: je 0xffffffff816e4af4 <bio_release_pages+15> <+6>: movzbl %sil,%esi <+10>: jmp 0xffffffff816dab7c <__bio_release_pages> <+15>: jmp 0xffffffff81d0b7c0 <__x86_return_thunk> instead. Also, the MOVZBL instruction looks unnecessary[2] - I think it's just 're-booling' the mark_dirty parameter. Signed-off-by: David Howells <dhowells@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: John Hubbard <jhubbard@nvidia.com> cc: Jens Axboe <axboe@kernel.dk> cc: linux-block@vger.kernel.org Link: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108370 [1] Link: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108371 [2] Link: https://lore.kernel.org/r/167391056756.2311931.356007731815807265.stgit@warthog.procyon.org.uk/ # v6 Reviewed-by: Christian Brauner <brauner@kernel.org> Reviewed-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20230522205744.2825689-3-dhowells@redhat.com Signed-off-by: Jens Axboe <axboe@kernel.dk> diff 320fb0f9 Sun Aug 28 20:22:37 MDT 2022 Yu Kuai <yukuai3@huawei.com> blk-throttle: fix that io throttle can only work for single bio Test scripts: cd /sys/fs/cgroup/blkio/ echo "8:0 1024" > blkio.throttle.write_bps_device echo $$ > cgroup.procs dd if=/dev/zero of=/dev/sda bs=10k count=1 oflag=direct & dd if=/dev/zero of=/dev/sda bs=10k count=1 oflag=direct & Test result: 10240 bytes (10 kB, 10 KiB) copied, 10.0134 s, 1.0 kB/s 10240 bytes (10 kB, 10 KiB) copied, 10.0135 s, 1.0 kB/s The problem is that the second bio is finished after 10s instead of 20s. Root cause: 1) second bio will be flagged: __blk_throtl_bio while (true) { ... if (sq->nr_queued[rw]) -> some bio is throttled already break }; bio_set_flag(bio, BIO_THROTTLED); -> flag the bio 2) flagged bio will be dispatched without waiting: throtl_dispatch_tg tg_may_dispatch tg_with_in_bps_limit if (bps_limit == U64_MAX || bio_flagged(bio, BIO_THROTTLED)) *wait = 0; -> wait time is zero return true; commit 9f5ede3c01f9 ("block: throttle split bio in case of iops limit") support to count split bios for iops limit, thus it adds flagged bio checking in tg_with_in_bps_limit() so that split bios will only count once for bps limit, however, it introduce a new problem that io throttle won't work if multiple bios are throttled. In order to fix the problem, handle iops/bps limit in different ways: 1) for iops limit, there is no flag to record if the bio is throttled, and iops is always applied. 2) for bps limit, original bio will be flagged with BIO_BPS_THROTTLED, and io throttle will ignore bio with the flag. Noted this patch also remove the code to set flag in __bio_clone(), it's introduced in commit 111be8839817 ("block-throttle: avoid double charge"), and author thinks split bio can be resubmited and throttled again, which is wrong because split bio will continue to dispatch from caller. Fixes: 9f5ede3c01f9 ("block: throttle split bio in case of iops limit") Cc: <stable@vger.kernel.org> Signed-off-by: Yu Kuai <yukuai3@huawei.com> Acked-by: Tejun Heo <tj@kernel.org> Link: https://lore.kernel.org/r/20220829022240.3348319-2-yukuai1@huaweicloud.com Signed-off-by: Jens Axboe <axboe@kernel.dk> diff 320fb0f9 Sun Aug 28 20:22:37 MDT 2022 Yu Kuai <yukuai3@huawei.com> blk-throttle: fix that io throttle can only work for single bio Test scripts: cd /sys/fs/cgroup/blkio/ echo "8:0 1024" > blkio.throttle.write_bps_device echo $$ > cgroup.procs dd if=/dev/zero of=/dev/sda bs=10k count=1 oflag=direct & dd if=/dev/zero of=/dev/sda bs=10k count=1 oflag=direct & Test result: 10240 bytes (10 kB, 10 KiB) copied, 10.0134 s, 1.0 kB/s 10240 bytes (10 kB, 10 KiB) copied, 10.0135 s, 1.0 kB/s The problem is that the second bio is finished after 10s instead of 20s. Root cause: 1) second bio will be flagged: __blk_throtl_bio while (true) { ... if (sq->nr_queued[rw]) -> some bio is throttled already break }; bio_set_flag(bio, BIO_THROTTLED); -> flag the bio 2) flagged bio will be dispatched without waiting: throtl_dispatch_tg tg_may_dispatch tg_with_in_bps_limit if (bps_limit == U64_MAX || bio_flagged(bio, BIO_THROTTLED)) *wait = 0; -> wait time is zero return true; commit 9f5ede3c01f9 ("block: throttle split bio in case of iops limit") support to count split bios for iops limit, thus it adds flagged bio checking in tg_with_in_bps_limit() so that split bios will only count once for bps limit, however, it introduce a new problem that io throttle won't work if multiple bios are throttled. In order to fix the problem, handle iops/bps limit in different ways: 1) for iops limit, there is no flag to record if the bio is throttled, and iops is always applied. 2) for bps limit, original bio will be flagged with BIO_BPS_THROTTLED, and io throttle will ignore bio with the flag. Noted this patch also remove the code to set flag in __bio_clone(), it's introduced in commit 111be8839817 ("block-throttle: avoid double charge"), and author thinks split bio can be resubmited and throttled again, which is wrong because split bio will continue to dispatch from caller. Fixes: 9f5ede3c01f9 ("block: throttle split bio in case of iops limit") Cc: <stable@vger.kernel.org> Signed-off-by: Yu Kuai <yukuai3@huawei.com> Acked-by: Tejun Heo <tj@kernel.org> Link: https://lore.kernel.org/r/20220829022240.3348319-2-yukuai1@huaweicloud.com Signed-off-by: Jens Axboe <axboe@kernel.dk> diff 320fb0f9 Sun Aug 28 20:22:37 MDT 2022 Yu Kuai <yukuai3@huawei.com> blk-throttle: fix that io throttle can only work for single bio Test scripts: cd /sys/fs/cgroup/blkio/ echo "8:0 1024" > blkio.throttle.write_bps_device echo $$ > cgroup.procs dd if=/dev/zero of=/dev/sda bs=10k count=1 oflag=direct & dd if=/dev/zero of=/dev/sda bs=10k count=1 oflag=direct & Test result: 10240 bytes (10 kB, 10 KiB) copied, 10.0134 s, 1.0 kB/s 10240 bytes (10 kB, 10 KiB) copied, 10.0135 s, 1.0 kB/s The problem is that the second bio is finished after 10s instead of 20s. Root cause: 1) second bio will be flagged: __blk_throtl_bio while (true) { ... if (sq->nr_queued[rw]) -> some bio is throttled already break }; bio_set_flag(bio, BIO_THROTTLED); -> flag the bio 2) flagged bio will be dispatched without waiting: throtl_dispatch_tg tg_may_dispatch tg_with_in_bps_limit if (bps_limit == U64_MAX || bio_flagged(bio, BIO_THROTTLED)) *wait = 0; -> wait time is zero return true; commit 9f5ede3c01f9 ("block: throttle split bio in case of iops limit") support to count split bios for iops limit, thus it adds flagged bio checking in tg_with_in_bps_limit() so that split bios will only count once for bps limit, however, it introduce a new problem that io throttle won't work if multiple bios are throttled. In order to fix the problem, handle iops/bps limit in different ways: 1) for iops limit, there is no flag to record if the bio is throttled, and iops is always applied. 2) for bps limit, original bio will be flagged with BIO_BPS_THROTTLED, and io throttle will ignore bio with the flag. Noted this patch also remove the code to set flag in __bio_clone(), it's introduced in commit 111be8839817 ("block-throttle: avoid double charge"), and author thinks split bio can be resubmited and throttled again, which is wrong because split bio will continue to dispatch from caller. Fixes: 9f5ede3c01f9 ("block: throttle split bio in case of iops limit") Cc: <stable@vger.kernel.org> Signed-off-by: Yu Kuai <yukuai3@huawei.com> Acked-by: Tejun Heo <tj@kernel.org> Link: https://lore.kernel.org/r/20220829022240.3348319-2-yukuai1@huaweicloud.com Signed-off-by: Jens Axboe <axboe@kernel.dk> diff 0df71650 Thu Mar 24 14:35:24 MDT 2022 Mike Snitzer <snitzer@kernel.org> block: allow using the per-cpu bio cache from bio_alloc_bioset Replace the BIO_PERCPU_CACHE bio-internal flag with a REQ_ALLOC_CACHE one that can be passed to bio_alloc / bio_alloc_bioset, and implement the percpu cache allocation logic in a helper called from bio_alloc_bioset. This allows any bio_alloc_bioset user to use the percpu caches instead of having the functionality tied to struct kiocb. Signed-off-by: Mike Snitzer <snitzer@kernel.org> [hch: refactored a bit] Signed-off-by: Christoph Hellwig <hch@lst.de> Link: https://lore.kernel.org/r/20220324203526.62306-2-snitzer@kernel.org Signed-off-by: Jens Axboe <axboe@kernel.dk> diff 85f5a74c Wed Apr 07 23:33:45 MDT 2021 Matthew Wilcox (Oracle) <willy@infradead.org> block: Add bio_add_folio() This is a thin wrapper around bio_add_page(). The main advantage here is the documentation that folios larger than 2GiB are not supported. It's not currently possible to allocate folios that large, but if it ever becomes possible, this function will fail gracefully instead of doing I/O to the wrong bytes. Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org> Reviewed-by: Jens Axboe <axboe@kernel.dk> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Darrick J. Wong <djwong@kernel.org> diff 1bb6b810 Wed Oct 27 06:21:07 MDT 2021 Pavel Begunkov <asml.silence@gmail.com> block: avoid extra iter advance with async iocb Nobody cares about iov iterators state if we return -EIOCBQUEUED, so as the we now have __blkdev_direct_IO_async(), which gets pages only once, we can skip expensive iov_iter_advance(). It's around 1-2% of all CPU spent. Signed-off-by: Pavel Begunkov <asml.silence@gmail.com> Link: https://lore.kernel.org/r/a6158edfbfa2ae3bc24aed29a72f035df18fad2f.1635337135.git.asml.silence@gmail.com Signed-off-by: Jens Axboe <axboe@kernel.dk> |
Completed in 208 milliseconds