- 21 May, 2016 40 commits
-
-
Matthew Wilcox authored
The subtraction was the wrong way round, leading to undefined behaviour (shift by an amount larger than the size of the type). Signed-off-by:
Matthew Wilcox <willy@linux.intel.com> Reviewed-by:
Ross Zwisler <ross.zwisler@linux.intel.com> Cc: Konstantin Khlebnikov <koct9i@gmail.com> Cc: Kirill Shutemov <kirill.shutemov@linux.intel.com> Cc: Jan Kara <jack@suse.com> Cc: Neil Brown <neilb@suse.de> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
Matthew Wilcox authored
The code I previously added to enable multiorder radix tree entries was untested and therefore buggy. This commit adds the support functions that Ross and I decided were necessary over a four-week period of iterating various designs. Signed-off-by:
-
Matthew Wilcox authored
I've been receiving increasingly concerned notes from 0day about how much my recent changes have been bloating the radix tree. Make it happier by only including multiorder support if CONFIG_TRANSPARENT_HUGEPAGES is set. This is an independent Kconfig option, so other radix tree users can also set it if they have a need. Signed-off-by:
-
Ross Zwisler authored
radix_tree_for_each_chunk() and radix_tree_for_each_chunk_slot() have never been used in the kernel since their introduction in 2012, so remove them. Signed-off-by:
-
Ross Zwisler authored
When we make changes to radix-tree.h in the regular kernel source (include/linux/radix-tree.h), we really want our test code to be rebuilt. We also include a few other headers from tools/include and probably want to rebuild if these have been changed. Update the makefile so that all of our objects will be rebuilt when any of the headers we depend on are changed. Signed-off-by:
-
Ross Zwisler authored
Currently the full suite of regression tests take upwards of 30 minutes to run on my development machine. The vast majority of this time is taken by the big_gang_check() and copy_tag_check() tests, which each run their tests through thousands of iterations...does this have value? Without big_gang_check() and copy_tag_check(), the test suite runs in around 15 seconds on my box. Honestly the first time I ever ran through the entire test suite was to gather the timings for this email - it simply takes too long to be useful on a normal basis. Instead, hide the excessive iterations through big_gang_check() and copy_tag_check() tests behind an '-l' flag (for "long run") in case they are still useful, but allow the regression test suite to complete in a reasonable amount of time. We still run each of these tests a few times (3 at present) to try and keep the test coverage. Signed-off-by:
-
Ross Zwisler authored
The defines in regression2.c are already in radix-tree.h and duplicating them in the test case makes experimenting with other values for the fan-out harder than necessary. Allow the user of the radix tree to decide what the fan-out should be rather than fixing it to 8 for non-kernel uses. Signed-off-by:
-
Matthew Wilcox authored
Fairly simple tests; add various items to the tree, then make sure we can find them again. Also check that a pointer that we know isn't in the tree is not found. Signed-off-by:
-
Matthew Wilcox authored
Add an empty linux/init.h, and definitions for a few parts of the kernel API either in use now, or to be used in the near future. Start using the common definitions in tools/include/linux, although more work needs to be done here. Signed-off-by:
-
Matthew Wilcox authored
Commit e6145236 ("radix_tree: add support for multi-order entries") left the impression that the support for multiorder radix tree entries was functional. As soon as Ross tried to use it, it became apparent that my testing was completely inadequate, and it didn't even work a little bit for orders that were not a multiple of shift. This series of patches is the result of about 6 weeks of redesign, reimplementation, testing, arguing and hair-pulling. The great news is that the test-suite is now far better than it was. That's reflected in the diffstat for the test-suite alone: 12 files changed, 436 insertions(+), 28 deletions(-) The highlight for users of the tree is that the restriction on the order of inserted entries being >= RADIX_TREE_MAP_SHIFT is now gone; the radix tree now supports any order between 0 and 64. For those who are interested in how the tree works, patch 9 is probably the most interesting one as it introduces the new machinery for handling sibling entries. I've tried to be fair in attributing authorship to the person who contributed the majority of the code in each patch; Ross has been an invaluable partner in the development of this support and it's fair to say that each of us has code in every commit. I should also express my appreciation of the 0day testing. It prompted me that I was bloating the tinyconfig in an unacceptable way, and it bisected to a commit which contained a rather nasty memory-corruption bug. This patch (of 29): The irqdomain code was checking for 0 or 1 entries, not 0 entries like the comment said they were. Introduce a new helper that will actually check for an empty tree. Signed-off-by:
Jan Kara <jack@suse.cz> Cc: Konstantin Khlebnikov <koct9i@gmail.com> Cc: Kirill Shutemov <kirill.shutemov@linux.intel.com> Cc: Neil Brown <neilb@suse.de> Signed-off-by:
-
Andy Shevchenko authored
Instead of opencoding let's use generic UUID library functions here. Signed-off-by:
Andy Shevchenko <andriy.shevchenko@linux.intel.com> Cc: Darren Hart <dvhart@linux.intel.com> Signed-off-by:
-
Andy Shevchenko authored
Instead of opencoding let's use generic UUID library functions here. Signed-off-by:
-
Andy Shevchenko authored
UUID library provides uuid_be type and uuid_be_to_bin() function. This substitutes open coded variant by generic library calls. Signed-off-by:
Matt Fleming <matt@codeblueprint.co.uk> Cc: Dmitry Kasatkin <dmitry.kasatkin@gmail.com> Cc: Mimi Zohar <zohar@linux.vnet.ibm.com> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk> Cc: Arnd Bergmann <arnd@arndb.de> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Jens Axboe <axboe@kernel.dk> Signed-off-by:
-
Andy Shevchenko authored
Instead of opencoding let's use generic UUID library functions here. Signed-off-by:
-
Andy Shevchenko authored
Generic UUID library defines structure type, macro to define UUID, and the length of the UUID string. This patch removes duplicate data structure definition, UUID string length constant as well as macro for UUID handling. Signed-off-by:
-
Andy Shevchenko authored
UUID library provides uuid_be type and uuid_be_to_bin() function. This substitutes open coded variant by generic library calls. Signed-off-by:
-
Andy Shevchenko authored
There is no point in keeping an address in the file since it's subject to change. While here, update Intel Copyright years. Signed-off-by:
-
Andy Shevchenko authored
There are new helpers in this patch: uuid_is_valid checks if a UUID is valid uuid_be_to_bin converts from string to binary (big endian) uuid_le_to_bin converts from string to binary (little endian) They will be used in future, i.e. in the following patches in the series. This also moves the indices arrays to lib/uuid.c to be shared accross modules. [andriy.shevchenko@linux.intel.com: fix typo] Signed-off-by:
-
Andy Shevchenko authored
Let's gather the UUID related functions under one hood. Signed-off-by:
-
Andy Shevchenko authored
Instead of open coded variant re-use extension that vsprintf.c provides us for ages. Signed-off-by:
-
Andy Shevchenko authored
There are few functions here and there along with type definitions that provide UUID API. This series consolidates everything under one hood and converts current users. This has been tested for a while internally, however it doesn't mean we covered all possible cases (especially accuracy of UUID constants after conversion). So, please test this as much as you can and provide your tag. We appreciate the effort. The ACPI conversion is postponed for now to sort more generic things out first. This patch (of 9): Since we have hex_byte_pack_upper() we may use it directly and avoid second loop. Signed-off-by:
-
Jiri Slaby authored
The MTA says: <yasutake.koichi@jp.panasonic.com>: unknown user: "yasutake.koichi@jp.panasonic.com" Link: http://lkml.kernel.org/r/1462776755-9607-1-git-send-email-jslaby@suse.czSigned-off-by:Jiri Slaby <jslaby@suse.cz> Cc: David Howells <dhowells@redhat.com> Signed-off-by:
-
Eric Engestrom authored
It looks like the email address for this mailing list doesn't exist anymore: <spear-devel@list.st.com>: host mxb-00178001.gslb.pphosted.com[91.207.212.93] said: 550 5.1.1 User Unknown (in reply to RCPT TO command) Signed-off-by:
Eric Engestrom <eric.engestrom@imgtec.com> Acked-by:
Viresh Kumar <viresh.kumar@linaro.org> Cc: "David S. Miller" <davem@davemloft.net> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Kalle Valo <kvalo@codeaurora.org> Cc: Mauro Carvalho Chehab <mchehab@osg.samsung.com> Cc: Guenter Roeck <linux@roeck-us.net> Cc: Jiri Slaby <jslaby@suse.com> Signed-off-by:
-
Jiri Slaby authored
$ host -t mx lists.openrisc.net Host lists.openrisc.net not found: 3(NXDOMAIN) Signed-off-by:
-
Petr Mladek authored
In NMI context, printk() messages are stored into per-CPU buffers to avoid a possible deadlock. They are normally flushed to the main ring buffer via an IRQ work. But the work is never called when the system calls panic() in the very same NMI handler. This patch tries to flush NMI buffers before the crash dump is generated. In this case it does not risk a double release and bails out when the logbuf_lock is already taken. The aim is to get the messages into the main ring buffer when possible. It makes them better accessible in the vmcore. Then the patch tries to flush the buffers second time when other CPUs are down. It might be more aggressive and reset logbuf_lock. The aim is to get the messages available for the consequent kmsg_dump() and console_flush_on_panic() calls. The patch causes vprintk_emit() to be called even in NMI context again. But it is done via printk_deferred() so that the console handling is skipped. Consoles use internal locks and we could not prevent a deadlock easily. They are explicitly called later when the crash dump is not generated, see console_flush_on_panic(). Signed-off-by:
Petr Mladek <pmladek@suse.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Daniel Thompson <daniel.thompson@linaro.org> Cc: David Miller <davem@davemloft.net> Cc: Ingo Molnar <mingo@redhat.com> Cc: Jan Kara <jack@suse.cz> Cc: Jiri Kosina <jkosina@suse.com> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Russell King <rmk+kernel@arm.linux.org.uk> Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by:
-
Petr Mladek authored
Testing has shown that the backtrace sometimes does not fit into the 4kB temporary buffer that is used in NMI context. The warnings are gone when I double the temporary buffer size. This patch doubles the buffer size and makes it configurable. Note that this problem existed even in the x86-specific implementation that was added by the commit a9edc880 ("x86/nmi: Perform a safe NMI stack trace on all CPUs"). Nobody noticed it because it did not print any warnings. Signed-off-by:
-
Petr Mladek authored
We could not resize the temporary buffer in NMI context. Let's warn if a message is lost. This is rather theoretical. printk() should not be used in NMI. The only sensible use is when we want to print backtrace from all CPUs. The current buffer should be enough for this purpose. [akpm@linux-foundation.org: whitespace fixlet] Signed-off-by:
-
Petr Mladek authored
printk() takes some locks and could not be used a safe way in NMI context. The chance of a deadlock is real especially when printing stacks from all CPUs. This particular problem has been addressed on x86 by the commit a9edc880 ("x86/nmi: Perform a safe NMI stack trace on all CPUs"). The patchset brings two big advantages. First, it makes the NMI backtraces safe on all architectures for free. Second, it makes all NMI messages almost safe on all architectures (the temporary buffer is limited. We still should keep the number of messages in NMI context at minimum). Note that there already are several messages printed in NMI context: WARN_ON(in_nmi()), BUG_ON(in_nmi()), anything being printed out from MCE handlers. These are not easy to avoid. This patch reuses most of the code and makes it generic. It is useful for all messages and architectures that support NMI. The alternative printk_func is set when entering and is reseted when leaving NMI context. It queues IRQ work to copy the messages into the main ring buffer in a safe context. __printk_nmi_flush() copies all available messages and reset the buffer. Then we could use a simple cmpxchg operations to get synchronized with writers. There is also used a spinlock to get synchronized with other flushers. We do not longer use seq_buf because it depends on external lock. It would be hard to make all supported operations safe for a lockless use. It would be confusing and error prone to make only some operations safe. The code is put into separate printk/nmi.c as suggested by Steven Rostedt. It needs a per-CPU buffer and is compiled only on architectures that call nmi_enter(). This is achieved by the new HAVE_NMI Kconfig flag. The are MN10300 and Xtensa architectures. We need to clean up NMI handling there first. Let's do it separately. The patch is heavily based on the draft from Peter Zijlstra, see https://lkml.org/lkml/2015/6/10/327 [arnd@arndb.de: printk-nmi: use %zu format string for size_t] [akpm@linux-foundation.org: min_t->min - all types are size_t here] Signed-off-by:
Peter Zijlstra <peterz@infradead.org> Suggested-by:
Steven Rostedt <rostedt@goodmis.org> Cc: Jan Kara <jack@suse.cz> Acked-by: Russell King <rmk+kernel@arm.linux.org.uk> [arm part] Cc: Daniel Thompson <daniel.thompson@linaro.org> Cc: Jiri Kosina <jkosina@suse.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: David Miller <davem@davemloft.net> Cc: Daniel Thompson <daniel.thompson@linaro.org> Signed-off-by:
-
René Nyffenegger authored
In include/linux/syscalls.h, the four functions sys_kill, sys_tgkill, sys_tkill and sys_rt_sigqueueinfo are declared with "int pid" and "int tgid". However, in kernel/signal.c, the corresponding definitions use the more appropriate "pid_t" (which is a typedef'd int). This patch changes "int" to "pid_t" in the declarations of sys_kill, sys_tgkill, sys_tkill and sys_rt_sigqueueinfo in <linux/syscalls.h> in order to harmonize the function declarations with their respective definitions. Link: http://lkml.kernel.org/r/57302FDA.7020205@renenyffenegger.chSigned-off-by:
René Nyffenegger <mail@renenyffenegger.ch> Cc: Andy Lutomirski <luto@amacapital.net> Cc: Josh Triplett <josh@joshtriplett.org> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: "Steven Rostedt (Red Hat)" <rostedt@goodmis.org> Cc: Zach Brown <zab@redhat.com> Cc: Milosz Tanski <milosz@adfin.com> Cc: Arnd Bergmann <arnd@arndb.de> Signed-off-by:
-
Jiri Slaby authored
When using this program (as root): #include <err.h> #include <stdio.h> #include <stdlib.h> #include <unistd.h> #include <sys/io.h> #include <sys/types.h> #include <sys/wait.h> #define ITER 1000 #define FORKERS 15 #define THREADS (6000/FORKERS) // 1850 is proc max static void fork_100_wait() { unsigned a, to_wait = 0; printf("\t%d forking %d\n", THREADS, getpid()); for (a = 0; a < THREADS; a++) { switch (fork()) { case 0: usleep(1000); exit(0); break; case -1: break; default: to_wait++; break; } } printf("\t%d forked from %d, waiting for %d\n", THREADS, getpid(), to_wait); for (a = 0; a < to_wait; a++) wait(NULL); printf("\t%d waited from %d\n", THREADS, getpid()); } static void run_forkers() { pid_t forkers[FORKERS]; unsigned a; for (a = 0; a < FORKERS; a++) { switch ((forkers[a] = fork())) { case 0: fork_100_wait(); exit(0); break; case -1: err(1, "DIE fork of %d'th forker", a); break; default: break; } } for (a = 0; a < FORKERS; a++) waitpid(forkers[a], NULL, 0); } int main() { unsigned a; int ret; ret = ioperm(10, 20, 0); if (ret < 0) err(1, "ioperm"); for (a = 0; a < ITER; a++) run_forkers(); return 0; } kmemleak reports many occurences of this leak: unreferenced object 0xffff8805917c8000 (size 8192): comm "fork-leak", pid 2932, jiffies 4295354292 (age 1871.028s) hex dump (first 32 bytes): ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ................ ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ................ backtrace: [<ffffffff814cfbf5>] kmemdup+0x25/0x50 [<ffffffff8103ab43>] copy_thread_tls+0x6c3/0x9a0 [<ffffffff81150174>] copy_process+0x1a84/0x5790 [<ffffffff811dc375>] wake_up_new_task+0x2d5/0x6f0 [<ffffffff8115411d>] _do_fork+0x12d/0x820 ... Due to the leakage of the memory items which should have been freed in arch/x86/kernel/process.c:exit_thread(). Make sure the memory is freed when fork fails later in copy_process. This is done by calling exit_thread with the thread to kill. Signed-off-by: -
Jiri Slaby authored
We need to call exit_thread from copy_process in a fail path. So make it accept task_struct as a parameter. [v2] * s390: exit_thread_runtime_instr doesn't make sense to be called for non-current tasks. * arm: fix the comment in vfp_thread_copy * change 'me' to 'tsk' for task_struct * now we can change only archs that actually have exit_thread [akpm@linux-foundation.org: coding-style fixes] Signed-off-by:
-
Jiri Slaby authored
Define HAVE_EXIT_THREAD for archs which want to do something in exit_thread. For others, let's define exit_thread as an empty inline. This is a cleanup before we change the prototype of exit_thread to accept a task parameter. [akpm@linux-foundation.org: fix mips] Signed-off-by:
-
Jiri Slaby authored
We need to call exit_thread from copy_process in a fail path. Since exit_thread on mn10300 calls exit_thread_runtime_instr, make it accept task_struct as a parameter now. Signed-off-by:
-
Janis Danisevskis authored
The PR_DUMPABLE flag causes the pid related paths of the proc file system to be owned by ROOT. The implementation of pthread_set/getname_np however needs access to /proc/<pid>/task/<tid>/comm. If PR_DUMPABLE is false this implementation is locked out. This patch installs a special permission function for the file "comm" that grants read and write access to all threads of the same group regardless of the ownership of the inode. For all other threads the function falls back to the generic inode permission check. [akpm@linux-foundation.org: fix spello in comment] Signed-off-by:
Janis Danisevskis <jdanis@google.com> Acked-by:
Kees Cook <keescook@chromium.org> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Cyrill Gorcunov <gorcunov@openvz.org> Cc: Alexey Dobriyan <adobriyan@gmail.com> Cc: Colin Ian King <colin.king@canonical.com> Cc: David Rientjes <rientjes@google.com> Cc: Minfei Huang <mnfhuang@gmail.com> Cc: John Stultz <john.stultz@linaro.org> Cc: Calvin Owens <calvinowens@fb.com> Cc: Jann Horn <jann@thejh.net> Signed-off-by:
-
Richard W.M. Jones authored
It's not possible to read the process umask without also modifying it, which is what umask(2) does. A library cannot read umask safely, especially if the main program might be multithreaded. Add a new status line ("Umask") in /proc/<PID>/status. It contains the file mode creation mask (umask) in octal. It is only shown for tasks which have task->fs. This patch is adapted from one originally written by Pierre Carrier. The use case is that we have endless trouble with people setting weird umask() values (usually on the grounds of "security"), and then everything breaking. I'm on the hook to fix these. We'd like to add debugging to our program so we can dump out the umask in debug reports. Previous versions of the patch used a syscall so you could only read your own umask. That's all I need. However there was quite a lot of push-back from those, so this new version exports it in /proc. See: https://lkml.org/lkml/2016/4/13/704 [umask2] https://lkml.org/lkml/2016/4/13/487 [getumask] Signed-off-by:Richard W.M. Jones <rjones@redhat.com> Acked-by:
Konstantin Khlebnikov <koct9i@gmail.com> Acked-by:
Jerome Marchand <jmarchan@redhat.com> Acked-by:
-
Sergey Senozhatsky authored
debug_stat sysfs is read-only and represents various debugging data that zram developers may need. This file is not meant to be used by anyone else: its content is not documented and will change any time w/o any notice. Therefore, the output of debug_stat file contains a version string. To avoid any confusion, we will increase the version number every time we modify the output. At the moment this file exports only one value -- the number of re-compressions, IOW, the number of times compression fast path has failed. This stat is temporary any will be useful in case if any per-cpu compression streams regressions will be reported. Link: http://lkml.kernel.org/r/20160513230834.GB26763@bbox Link: http://lkml.kernel.org/r/20160511134553.12655-1-sergey.senozhatsky@gmail.comSigned-off-by:
Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by:
Minchan Kim <minchan@kernel.org> Signed-off-by:
-
Sergey Senozhatsky authored
Remove the internal part of max_comp_streams interface, since we switched to per-cpu streams. We will keep RW max_comp_streams attr around, because: a) we may (silently) switch back to idle compression streams list and don't want to disturb user space b) max_comp_streams attr must wait for the next 'lay off cycle'; we give user space 2 years to adjust before we remove/downgrade the attr, and there are already several attrs scheduled for removal in 4.11, so it's too late for max_comp_streams. This slightly change a user visible behaviour: - First, reading from max_comp_stream file now will always return the number of online CPUs. - Second, writing to max_comp_stream will not take any effect. Link: http://lkml.kernel.org/r/20160503165546.25201-1-sergey.senozhatsky@gmail.comSigned-off-by:
-
Dan Streetman authored
Change the return type of zs_pool_stat_create() to void, and remove the logic to abort pool creation if the stat debugfs dir/file could not be created. The debugfs stat file is for debugging/information only, and doesn't affect operation of zsmalloc; there is no reason to abort creating the pool if the stat file can't be created. This was seen with zswap, which used the same name for all pool creations, which caused zsmalloc to fail to create a second pool for zswap if CONFIG_ZSMALLOC_STAT was enabled. Signed-off-by:
Dan Streetman <ddstreet@ieee.org> Reviewed-by:
-
Dan Streetman authored
Add a work_struct to struct zswap_pool, and change __zswap_pool_empty to use the workqueue instead of using call_rcu(). When zswap destroys a pool no longer in use, it uses call_rcu() to perform the destruction/freeing. Since that executes in softirq context, it must not sleep. However, actually destroying the pool involves freeing the per-cpu compressors (which requires locking the cpu_add_remove_lock mutex) and freeing the zpool, for which the implementation may sleep (e.g. zsmalloc calls kmem_cache_destroy, which locks the slab_mutex). So if either mutex is currently taken, or any other part of the compressor or zpool implementation sleeps, it will result in a BUG(). It's not easy to reproduce this when changing zswap's params normally. In testing with a loaded system, this does not fail: $ cd /sys/module/zswap/parameters $ echo lz4 > compressor ; echo zsmalloc > zpool nor does this: $ while true ; do > echo lzo > compressor ; echo zbud > zpool > sleep 1 > echo lz4 > compressor ; echo zsmalloc > zpool > sleep 1 > done although it's still possible either of those might fail, depending on whether anything else besides zswap has locked the mutexes. However, changing a parameter with no delay immediately causes the schedule while atomic BUG: $ while true ; do > echo lzo > compressor ; echo lz4 > compressor > done This is essentially the same as Yu Zhao's proposed patch to zsmalloc, but moved to zswap, to cover compressor and zpool freeing. Fixes: f1c54846 ("zswap: dynamic pool creation") Signed-off-by:
Yu Zhao <yuzhao@google.com> Reviewed-by:
-
Sergey Senozhatsky authored
Remove idle streams list and keep compression streams in per-cpu data. This removes two contented spin_lock()/spin_unlock() calls from write path and also prevent write OP from being preempted while holding the compression stream, which can cause slow downs. For instance, let's assume that we have N cpus and N-2 max_comp_streams.TASK1 owns the last idle stream, TASK2-TASK3 come in with the write requests: TASK1 TASK2 TASK3 zram_bvec_write() spin_lock find stream spin_unlock compress <<preempted>> zram_bvec_write() spin_lock find stream spin_unlock no_stream schedule zram_bvec_write() spin_lock find_stream spin_unlock no_stream schedule spin_lock release stream spin_unlock wake up TASK2 not only TASK2 and TASK3 will not get the stream, TASK1 will be preempted in the middle of its operation; while we would prefer it to finish compression and release the stream. Test environment: x86_64, 4 CPU box, 3G zram, lzo The following fio tests were executed: read, randread, write, randwrite, rw, randrw with the increasing number of jobs from 1 to 10. 4 streams 8 streams per-cpu =========================================================== jobs1 READ: 2520.1MB/s 2566.5MB/s 2491.5MB/s READ: 2102.7MB/s 2104.2MB/s 2091.3MB/s WRITE: 1355.1MB/s 1320.2MB/s 1378.9MB/s WRITE: 1103.5MB/s 1097.2MB/s 1122.5MB/s READ: 434013KB/s 435153KB/s 439961KB/s WRITE: 433969KB/s 435109KB/s 439917KB/s READ: 403166KB/s 405139KB/s 403373KB/s WRITE: 403223KB/s 405197KB/s 403430KB/s jobs2 READ: 7958.6MB/s 8105.6MB/s 8073.7MB/s READ: 6864.9MB/s 6989.8MB/s 7021.8MB/s WRITE: 2438.1MB/s 2346.9MB/s 3400.2MB/s WRITE: 1994.2MB/s 1990.3MB/s 2941.2MB/s READ: 981504KB/s 973906KB/s 1018.8MB/s WRITE: 981659KB/s 974060KB/s 1018.1MB/s READ: 937021KB/s 938976KB/s 987250KB/s WRITE: 934878KB/s 936830KB/s 984993KB/s jobs3 READ: 13280MB/s 13553MB/s 13553MB/s READ: 11534MB/s 11785MB/s 11755MB/s WRITE: 3456.9MB/s 3469.9MB/s 4810.3MB/s WRITE: 3029.6MB/s 3031.6MB/s 4264.8MB/s READ: 1363.8MB/s 1362.6MB/s 1448.9MB/s WRITE: 1361.9MB/s 1360.7MB/s 1446.9MB/s READ: 1309.4MB/s 1310.6MB/s 1397.5MB/s WRITE: 1307.4MB/s 1308.5MB/s 1395.3MB/s jobs4 READ: 20244MB/s 20177MB/s 20344MB/s READ: 17886MB/s 17913MB/s 17835MB/s WRITE: 4071.6MB/s 4046.1MB/s 6370.2MB/s WRITE: 3608.9MB/s 3576.3MB/s 5785.4MB/s READ: 1824.3MB/s 1821.6MB/s 1997.5MB/s WRITE: 1819.8MB/s 1817.4MB/s 1992.5MB/s READ: 1765.7MB/s 1768.3MB/s 1937.3MB/s WRITE: 1767.5MB/s 1769.1MB/s 1939.2MB/s jobs5 READ: 18663MB/s 18986MB/s 18823MB/s READ: 16659MB/s 16605MB/s 16954MB/s WRITE: 3912.4MB/s 3888.7MB/s 6126.9MB/s WRITE: 3506.4MB/s 3442.5MB/s 5519.3MB/s READ: 1798.2MB/s 1746.5MB/s 1935.8MB/s WRITE: 1792.7MB/s 1740.7MB/s 1929.1MB/s READ: 1727.6MB/s 1658.2MB/s 1917.3MB/s WRITE: 1726.5MB/s 1657.2MB/s 1916.6MB/s jobs6 READ: 21017MB/s 20922MB/s 21162MB/s READ: 19022MB/s 19140MB/s 18770MB/s WRITE: 3968.2MB/s 4037.7MB/s 6620.8MB/s WRITE: 3643.5MB/s 3590.2MB/s 6027.5MB/s READ: 1871.8MB/s 1880.5MB/s 2049.9MB/s WRITE: 1867.8MB/s 1877.2MB/s 2046.2MB/s READ: 1755.8MB/s 1710.3MB/s 1964.7MB/s WRITE: 1750.5MB/s 1705.9MB/s 1958.8MB/s jobs7 READ: 21103MB/s 20677MB/s 21482MB/s READ: 18522MB/s 18379MB/s 19443MB/s WRITE: 4022.5MB/s 4067.4MB/s 6755.9MB/s WRITE: 3691.7MB/s 3695.5MB/s 5925.6MB/s READ: 1841.5MB/s 1933.9MB/s 2090.5MB/s WRITE: 1842.7MB/s 1935.3MB/s 2091.9MB/s READ: 1832.4MB/s 1856.4MB/s 1971.5MB/s WRITE: 1822.3MB/s 1846.2MB/s 1960.6MB/s jobs8 READ: 20463MB/s 20194MB/s 20862MB/s READ: 18178MB/s 17978MB/s 18299MB/s WRITE: 4085.9MB/s 4060.2MB/s 7023.8MB/s WRITE: 3776.3MB/s 3737.9MB/s 6278.2MB/s READ: 1957.6MB/s 1944.4MB/s 2109.5MB/s WRITE: 1959.2MB/s 1946.2MB/s 2111.4MB/s READ: 1900.6MB/s 1885.7MB/s 2082.1MB/s WRITE: 1896.2MB/s 1881.4MB/s 2078.3MB/s jobs9 READ: 19692MB/s 19734MB/s 19334MB/s READ: 17678MB/s 18249MB/s 17666MB/s WRITE: 4004.7MB/s 4064.8MB/s 6990.7MB/s WRITE: 3724.7MB/s 3772.1MB/s 6193.6MB/s READ: 1953.7MB/s 1967.3MB/s 2105.6MB/s WRITE: 1953.4MB/s 1966.7MB/s 2104.1MB/s READ: 1860.4MB/s 1897.4MB/s 2068.5MB/s WRITE: 1858.9MB/s 1895.9MB/s 2066.8MB/s jobs10 READ: 19730MB/s 19579MB/s 19492MB/s READ: 18028MB/s 18018MB/s 18221MB/s WRITE: 4027.3MB/s 4090.6MB/s 7020.1MB/s WRITE: 3810.5MB/s 3846.8MB/s 6426.8MB/s READ: 1956.1MB/s 1994.6MB/s 2145.2MB/s WRITE: 1955.9MB/s 1993.5MB/s 2144.8MB/s READ: 1852.8MB/s 1911.6MB/s 2075.8MB/s WRITE: 1855.7MB/s 1914.6MB/s 2078.1MB/s perf stat 4 streams 8 streams per-cpu ==================================================================================================================== jobs1 stalled-cycles-frontend 23,174,811,209 ( 38.21%) 23,220,254,188 ( 38.25%) 23,061,406,918 ( 38.34%) stalled-cycles-backend 11,514,174,638 ( 18.98%) 11,696,722,657 ( 19.27%) 11,370,852,810 ( 18.90%) instructions 73,925,005,782 ( 1.22) 73,903,177,632 ( 1.22) 73,507,201,037 ( 1.22) branches 14,455,124,835 ( 756.063) 14,455,184,779 ( 755.281) 14,378,599,509 ( 758.546) branch-misses 69,801,336 ( 0.48%) 80,225,529 ( 0.55%) 72,044,726 ( 0.50%) jobs2 stalled-cycles-frontend 49,912,741,782 ( 46.11%) 50,101,189,290 ( 45.95%) 32,874,195,633 ( 35.11%) stalled-cycles-backend 27,080,366,230 ( 25.02%) 27,949,970,232 ( 25.63%) 16,461,222,706 ( 17.58%) instructions 122,831,629,690 ( 1.13) 122,919,846,419 ( 1.13) 121,924,786,775 ( 1.30) branches 23,725,889,239 ( 692.663) 23,733,547,140 ( 688.062) 23,553,950,311 ( 794.794) branch-misses 90,733,041 ( 0.38%) 96,320,895 ( 0.41%) 84,561,092 ( 0.36%) jobs3 stalled-cycles-frontend 66,437,834,608 ( 45.58%) 63,534,923,344 ( 43.69%) 42,101,478,505 ( 33.19%) stalled-cycles-backend 34,940,799,661 ( 23.97%) 34,774,043,148 ( 23.91%) 21,163,324,388 ( 16.68%) instructions 171,692,121,862 ( 1.18) 171,775,373,044 ( 1.18) 170,353,542,261 ( 1.34) branches 32,968,962,622 ( 628.723) 32,987,739,894 ( 630.512) 32,729,463,918 ( 717.027) branch-misses 111,522,732 ( 0.34%) 110,472,894 ( 0.33%) 99,791,291 ( 0.30%) jobs4 stalled-cycles-frontend 98,741,701,675 ( 49.72%) 94,797,349,965 ( 47.59%) 54,535,655,381 ( 33.53%) stalled-cycles-backend 54,642,609,615 ( 27.51%) 55,233,554,408 ( 27.73%) 27,882,323,541 ( 17.14%) instructions 220,884,807,851 ( 1.11) 220,930,887,273 ( 1.11) 218,926,845,851 ( 1.35) branches 42,354,518,180 ( 592.105) 42,362,770,587 ( 590.452) 41,955,552,870 ( 716.154) branch-misses 138,093,449 ( 0.33%) 131,295,286 ( 0.31%) 121,794,771 ( 0.29%) jobs5 stalled-cycles-frontend 116,219,747,212 ( 48.14%) 110,310,397,012 ( 46.29%) 66,373,082,723 ( 33.70%) stalled-cycles-backend 66,325,434,776 ( 27.48%) 64,157,087,914 ( 26.92%) 32,999,097,299 ( 16.76%) instructions 270,615,008,466 ( 1.12) 270,546,409,525 ( 1.14) 268,439,910,948 ( 1.36) branches 51,834,046,557 ( 599.108) 51,811,867,722 ( 608.883) 51,412,576,077 ( 729.213) branch-misses 158,197,086 ( 0.31%) 142,639,805 ( 0.28%) 133,425,455 ( 0.26%) jobs6 stalled-cycles-frontend 138,009,414,492 ( 48.23%) 139,063,571,254 ( 48.80%) 75,278,568,278 ( 32.80%) stalled-cycles-backend 79,211,949,650 ( 27.68%) 79,077,241,028 ( 27.75%) 37,735,797,899 ( 16.44%) instructions 319,763,993,731 ( 1.12) 319,937,782,834 ( 1.12) 316,663,600,784 ( 1.38) branches 61,219,433,294 ( 595.056) 61,250,355,540 ( 598.215) 60,523,446,617 ( 733.706) branch-misses 169,257,123 ( 0.28%) 154,898,028 ( 0.25%) 141,180,587 ( 0.23%) jobs7 stalled-cycles-frontend 162,974,812,119 ( 49.20%) 159,290,061,987 ( 48.43%) 88,046,641,169 ( 33.21%) stalled-cycles-backend 92,223,151,661 ( 27.84%) 91,667,904,406 ( 27.87%) 44,068,454,971 ( 16.62%) instructions 369,516,432,430 ( 1.12) 369,361,799,063 ( 1.12) 365,290,380,661 ( 1.38) branches 70,795,673,950 ( 594.220) 70,743,136,124 ( 597.876) 69,803,996,038 ( 732.822) branch-misses 181,708,327 ( 0.26%) 165,767,821 ( 0.23%) 150,109,797 ( 0.22%) jobs8 stalled-cycles-frontend 185,000,017,027 ( 49.30%) 182,334,345,473 ( 48.37%) 99,980,147,041 ( 33.26%) stalled-cycles-backend 105,753,516,186 ( 28.18%) 107,937,830,322 ( 28.63%) 51,404,177,181 ( 17.10%) instructions 418,153,161,055 ( 1.11) 418,308,565,828 ( 1.11) 413,653,475,581 ( 1.38) branches 80,035,882,398 ( 592.296) 80,063,204,510 ( 589.843) 79,024,105,589 ( 730.530) branch-misses 199,764,528 ( 0.25%) 177,936,926 ( 0.22%) 160,525,449 ( 0.20%) jobs9 stalled-cycles-frontend 210,941,799,094 ( 49.63%) 204,714,679,254 ( 48.55%) 114,251,113,756 ( 33.96%) stalled-cycles-backend 122,640,849,067 ( 28.85%) 122,188,553,256 ( 28.98%) 58,360,041,127 ( 17.35%) instructions 468,151,025,415 ( 1.10) 467,354,869,323 ( 1.11) 462,665,165,216 ( 1.38) branches 89,657,067,510 ( 585.628) 89,411,550,407 ( 588.990) 88,360,523,943 ( 730.151) branch-misses 218,292,301 ( 0.24%) 191,701,247 ( 0.21%) 178,535,678 ( 0.20%) jobs10 stalled-cycles-frontend 233,595,958,008 ( 49.81%) 227,540,615,689 ( 49.11%) 160,341,979,938 ( 43.07%) stalled-cycles-backend 136,153,676,021 ( 29.03%) 133,635,240,742 ( 28.84%) 65,909,135,465 ( 17.70%) instructions 517,001,168,497 ( 1.10) 516,210,976,158 ( 1.11) 511,374,038,613 ( 1.37) branches 98,911,641,329 ( 585.796) 98,700,069,712 ( 591.583) 97,646,761,028 ( 728.712) branch-misses 232,341,823 ( 0.23%) 199,256,308 ( 0.20%) 183,135,268 ( 0.19%) per-cpu streams tend to cause significantly less stalled cycles; execute less branches and hit less branch-misses. perf stat reported execution time 4 streams 8 streams per-cpu ==================================================================== jobs1 seconds elapsed 20.909073870 20.875670495 20.817838540 jobs2 seconds elapsed 18.529488399 18.720566469 16.356103108 jobs3 seconds elapsed 18.991159531 18.991340812 16.766216066 jobs4 seconds elapsed 19.560643828 19.551323547 16.246621715 jobs5 seconds elapsed 24.746498464 25.221646740 20.696112444 jobs6 seconds elapsed 28.258181828 28.289765505 22.885688857 jobs7 seconds elapsed 32.632490241 31.909125381 26.272753738 jobs8 seconds elapsed 35.651403851 36.027596308 29.108024711 jobs9 seconds elapsed 40.569362365 40.024227989 32.898204012 jobs10 seconds elapsed 44.673112304 43.874898137 35.632952191 Please see Link: http://marc.info/?l=linux-kernel&m=146166970727530 Link: http://marc.info/?l=linux-kernel&m=146174716719650 for more test results (under low memory conditions). Signed-off-by:
-
