- 17 Apr, 2015 40 commits
-
-
Andrey Ryabinin authored
gcov profiling if enabled with other heavy compile-time instrumentation like KASan could trigger following softlockups: NMI watchdog: BUG: soft lockup - CPU#0 stuck for 22s! [swapper/0:1] Modules linked in: irq event stamp: 22823276 hardirqs last enabled at (22823275): [<ffffffff86e8d10d>] mutex_lock_nested+0x7d9/0x930 hardirqs last disabled at (22823276): [<ffffffff86e9521d>] apic_timer_interrupt+0x6d/0x80 softirqs last enabled at (22823172): [<ffffffff811ed969>] __do_softirq+0x4db/0x729 softirqs last disabled at (22823167): [<ffffffff811edfcf>] irq_exit+0x7d/0x15b CPU: 0 PID: 1 Comm: swapper/0 Tainted: G W 3.19.0-05245-gbb33326-dirty #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.7.5.1-0-g8936dbb-20141113_115728-nilsson.home.kraxel.org 04/01/2014 task: ffff88006cba8000 ti: ffff88006cbb0000 task.ti: ffff88006cbb0000 RIP: kasan_mem_to_shadow+0x1e/0x1f Call Trace: strcmp+0x28/0x70 get_node_by_name+0x66/0x99 gcov_event+0x4f/0x69e gcov_enable_events+0x54/0x7b gcov_fs_init+0xf8/0x134 do_one_initcall+0x1b2/0x288 kernel_init_freeable+0x467/0x580 kernel_init+0x15/0x18b ret_from_fork+0x7c/0xb0 Kernel panic - not syncing: softlockup: hung tasks Fix this by sticking cond_resched() in gcov_enable_events(). Signed-off-by:Andrey Ryabinin <a.ryabinin@samsung.com> Reported-by:
Fengguang Wu <fengguang.wu@intel.com> Cc: Peter Oberparleiter <oberpar@linux.vnet.ibm.com> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
Heinrich Schuchardt authored
When converting unsigned long to int overflows may occur. These currently are not detected when writing to the sysctl file system. E.g. on a system where int has 32 bits and long has 64 bits echo 0x800001234 > /proc/sys/kernel/threads-max has the same effect as echo 0x1234 > /proc/sys/kernel/threads-max The patch adds the missing check in do_proc_dointvec_conv. With the patch an overflow will result in an error EINVAL when writing to the the sysctl file system. Signed-off-by:
Heinrich Schuchardt <xypron.glpk@gmx.de> Signed-off-by:
-
Sergey Senozhatsky authored
cpumask_next_and() is looking for cpumask_next() in src1 in a loop and tests if found cpu is also present in src2. remove that loop, perform cpumask_and() of src1 and src2 first and use that new mask to find cpumask_next(). Apart from removing while loop, ./bloat-o-meter on x86_64 shows add/remove: 0/0 grow/shrink: 0/1 up/down: 0/-8 (-8) function old new delta cpumask_next_and 62 54 -8 Signed-off-by:
Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Cc: Tejun Heo <tj@kernel.org> Cc: "David S. Miller" <davem@davemloft.net> Cc: Amir Vadai <amirv@mellanox.com> Signed-off-by:
-
Kirill Tkhai authored
We set sig->notify_count = -1 between RELEASE and ACQUIRE operations: spin_unlock_irq(lock); ... if (!thread_group_leader(tsk)) { ... for (;;) { sig->notify_count = -1; write_lock_irq(&tasklist_lock); There are no restriction on it so other processors may see this STORE mixed with other STOREs in both areas limited by the spinlocks. Probably, it may be reordered with the above sig->group_exit_task = tsk; sig->notify_count = zap_other_threads(tsk); in some way. Set it under tasklist_lock locked to be sure nothing will be reordered. Signed-off-by:Kirill Tkhai <ktkhai@parallels.com> Acked-by:
Oleg Nesterov <oleg@redhat.com> Signed-off-by:
-
Davidlohr Bueso authored
Oleg cleverly suggested using xchg() to set the new mm->exe_file instead of calling set_mm_exe_file() which requires some form of serialization -- mmap_sem in this case. For archs that do not have atomic rmw instructions we still fallback to a spinlock alternative, so this should always be safe. As such, we only need the mmap_sem for looking up the backing vm_file, which can be done sharing the lock. Naturally, this means we need to manually deal with both the new and old file reference counting, and we need not worry about the MMF_EXE_FILE_CHANGED bits, which can probably be deleted in the future anyway. Signed-off-by:
Davidlohr Bueso <dbueso@suse.de> Suggested-by:
Konstantin Khlebnikov <khlebnikov@yandex-team.ru> Signed-off-by:
-
Konstantin Khlebnikov authored
This patch removes mm->mmap_sem from mm->exe_file read side. Also it kills dup_mm_exe_file() and moves exe_file duplication into dup_mmap() where both mmap_sems are locked. [akpm@linux-foundation.org: fix comment typo] Signed-off-by:
-
Heinrich Schuchardt authored
File /proc/sys/kernel/threads-max controls the maximum number of threads that can be created using fork(). [akpm@linux-foundation.org: fix typo, per Guenter] Signed-off-by:
-
Heinrich Schuchardt authored
Users can change the maximum number of threads by writing to /proc/sys/kernel/threads-max. With the patch the value entered is checked against the same limits that apply when fork_init is called. Signed-off-by:
-
Heinrich Schuchardt authored
PAGE_SIZE is not guaranteed to be equal to or less than 8 times the THREAD_SIZE. E.g. architecture hexagon may have page size 1M and thread size 4096. This would lead to a division by zero in the calculation of max_threads. With 32-bit calculation there is no solution which delivers valid results for all possible combinations of the parameters. The code is only called once. Hence a 64-bit calculation can be used as solution. [akpm@linux-foundation.org: use clamp_t(), per Oleg] Signed-off-by:
-
Heinrich Schuchardt authored
PAGE_SIZE is not guaranteed to be equal to or less than 8 times the THREAD_SIZE. E.g. architecture hexagon may have page size 1M and thread size 4096. This would lead to a division by zero in the calculation of max_threads. With this patch the buggy code is moved to a separate function set_max_threads. The error is not fixed. After fixing the problem in a separate patch the new function can be reused to adjust max_threads after adding or removing memory. Argument mempages of function fork_init() is removed as totalram_pages is an exported symbol. The creation of separate patches for refactoring to a new function and for fixing the logic was suggested by Ingo Molnar. Signed-off-by:
-
Jean Delvare authored
The comment explaining what value max_threads is set to is outdated. The maximum memory consumption ratio for thread structures was 1/2 until February 2002, then it was briefly changed to 1/16 before being set to 1/8 which we still use today. The comment was never updated to reflect that change, it's about time. Signed-off-by:
Jean Delvare <jdelvare@suse.de> Signed-off-by:
-
Michal Hocko authored
copy_process will report any failure in alloc_pid as ENOMEM currently which is misleading because the pid allocation might fail not only when the memory is short but also when the pid space is consumed already. The current man page even mentions this case: : EAGAIN : : A system-imposed limit on the number of threads was encountered. : There are a number of limits that may trigger this error: the : RLIMIT_NPROC soft resource limit (set via setrlimit(2)), which : limits the number of processes and threads for a real user ID, was : reached; the kernel's system-wide limit on the number of processes : and threads, /proc/sys/kernel/threads-max, was reached (see : proc(5)); or the maximum number of PIDs, /proc/sys/kernel/pid_max, : was reached (see proc(5)). so the current behavior is also incorrect wrt. documentation. POSIX man page also suggest returing EAGAIN when the process count limit is reached. This patch simply propagates error code from alloc_pid and makes sure we return -EAGAIN due to reservation failure. This will make behavior of fork closer to both our documentation and POSIX. alloc_pid might alsoo fail when the reaper in the pid namespace is dead (the namespace basically disallows all new processes) and there is no good error code which would match documented ones. We have traditionally returned ENOMEM for this case which is misleading as well but as per Eric W. Biederman this behavior is documented in man pid_namespaces(7) : If the "init" process of a PID namespace terminates, the kernel : terminates all of the processes in the namespace via a SIGKILL signal. : This behavior reflects the fact that the "init" process is essential for : the correct operation of a PID namespace. In this case, a subsequent : fork(2) into this PID namespace will fail with the error ENOMEM; it is : not possible to create a new processes in a PID namespace whose "init" : process has terminated. and introducing a new error code would be too risky so let's stick to ENOMEM for this case. Signed-off-by:
Michal Hocko <mhocko@suse.cz> Cc: Oleg Nesterov <oleg@redhat.com> Cc: "Eric W. Biederman" <ebiederm@xmission.com> Cc: Michael Kerrisk <mtk.manpages@gmail.com> Signed-off-by:
-
Vladimir Davydov authored
Sending SI_TKILL from rt_[tg]sigqueueinfo was deprecated, so now we issue a warning on the first attempt of doing it. We use WARN_ON_ONCE, which is not informative and, what is worse, taints the kernel, making the trinity syscall fuzzer complain false-positively from time to time. It does not look like we need this warning at all, because the behaviour changed quite a long time ago (2.6.39), and if an application relies on the old API, it gets EPERM anyway and can issue a warning by itself. So let us zap the warning in kernel. Signed-off-by:
Vladimir Davydov <vdavydov@parallels.com> Acked-by:
-
Oleg Nesterov authored
ptrace_detach() re-checks ->ptrace under tasklist lock and calls release_task() if __ptrace_detach() returns true. This was needed because the __TASK_TRACED tracee could be killed/untraced, and it could even pass exit_notify() before we take tasklist_lock. But this is no longer possible after 9899d11f "ptrace: ensure arch_ptrace/ptrace_request can never race with SIGKILL". We can turn these checks into WARN_ON() and remove release_task(). While at it, document the setting of child->exit_code. Signed-off-by:
-
Oleg Nesterov authored
ptrace_resume() is called when the tracee is still __TASK_TRACED. We set tracee->exit_code and then wake_up_state() changes tracee->state. If the tracer's sub-thread does wait() in between, task_stopped_code(ptrace => T) wrongly looks like another report from tracee. This confuses debugger, and since wait_task_stopped() clears ->exit_code the tracee can miss a signal. Test-case: #include <stdio.h> #include <unistd.h> #include <sys/wait.h> #include <sys/ptrace.h> #include <pthread.h> #include <assert.h> int pid; void *waiter(void *arg) { int stat; for (;;) { assert(pid == wait(&stat)); assert(WIFSTOPPED(stat)); if (WSTOPSIG(stat) == SIGHUP) continue; assert(WSTOPSIG(stat) == SIGCONT); printf("ERR! extra/wrong report:%x\n", stat); } } int main(void) { pthread_t thread; pid = fork(); if (!pid) { assert(ptrace(PTRACE_TRACEME, 0,0,0) == 0); for (;;) kill(getpid(), SIGHUP); } assert(pthread_create(&thread, NULL, waiter, NULL) == 0); for (;;) ptrace(PTRACE_CONT, pid, 0, SIGCONT); return 0; } Note for stable: the bug is very old, but without 9899d11f "ptrace: ensure arch_ptrace/ptrace_request can never race with SIGKILL" the fix should use lock_task_sighand(child). Signed-off-by:Pavel Labath <labath@google.com> Tested-by:
-
Alexander Kuleshov authored
Signed-off-by:
Alexander Kuleshov <kuleshovmail@gmail.com> Acked-by:
OGAWA Hirofumi <hirofumi@mail.parknet.co.jp> Signed-off-by:
-
Alexander Kuleshov authored
'fat.h' includes <linux/buffer_head.h> which includes <linux/fs.h> which includes all the header files required for all *.c files fat filesystem. [akpm@linux-foundation.org: fs/fat/iode.c needs seq_file.h] [sfr@canb.auug.org.au: put one actually necessary include file back] Signed-off-by:
Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by:
-
Alexander Kuleshov authored
'*sb' never used, so let's remote it and pass inode->i_sb directly to the MSDOS_SB. Signed-off-by:
-
Thomas Hebb authored
On Mac OS X, HFS+ extended attributes are not namespaced. Since we want to be compatible with OS X filesystems and yet still support the Linux namespacing system, the hfsplus driver implements a special "osx" namespace that is reported for any attribute that is not namespaced on-disk. However, the current code for getting and setting these unprefixed attributes is broken. hfsplus_osx_setattr() and hfsplus_osx_getattr() are passed names that have already had their "osx." prefixes stripped by the generic functions. The functions first, quite correctly, check those names to make sure that they aren't prefixed with a known namespace, which would allow namespace access restrictions to be bypassed. However, the functions then prepend "osx." to the name they're given before passing it on to hfsplus_getattr() and hfsplus_setattr(). Not only does this cause the "osx." prefix to be stored on-disk, defeating its purpose, it also breaks the check for the special "com.apple.FinderInfo" attribute, which is reported for all files, and as a consequence makes some userspace applications (e.g. GNU patch) fail even when extended attributes are not otherwise in use. There are five commits which have touched this particular code: 127e5f5a ("hfsplus: rework functionality of getting, setting and deleting of extended attributes") b168fff7 ("hfsplus: use xattr handlers for removexattr") bf29e886 ("hfsplus: correct usage of HFSPLUS_ATTR_MAX_STRLEN for non-English attributes") fcacbd95e121 ("fs/hfsplus: move xattr_name allocation in hfsplus_getxattr()") ec1bbd346f18 ("fs/hfsplus: move xattr_name allocation in hfsplus_setxattr()") The first commit creates the functions to begin with. The namespace is prepended by the original code, which I believe was correct at the time, since hfsplus_?etattr() stripped the prefix if found. The second commit removes this behavior from hfsplus_?etattr() and appears to have been intended to also remove the prefixing from hfsplus_osx_?etattr(). However, what it actually does is remove a necessary strncpy() call completely, breaking the osx namespace entirely. The third commit re-adds the strncpy() call as it was originally, but doesn't mention it in its commit message. The final two commits refactor the code and don't affect its functionality. This commit does what b168fff7 attempted to do (prevent the prefix from being added), but does it properly, instead of passing in an empty buffer (which is what b168fff7 actually did). Fixes: b168fff7 ("hfsplus: use xattr handlers for removexattr") Signed-off-by:
Thomas Hebb <tommyhebb@gmail.com> Cc: Hin-Tak Leung <htl10@users.sourceforge.net> Cc: Sergei Antonov <saproj@gmail.com> Cc: Anton Altaparmakov <anton@tuxera.com> Cc: Fabian Frederick <fabf@skynet.be> Cc: Christian Kujau <lists@nerdbynature.de> Cc: Christoph Hellwig <hch@infradead.org> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Viacheslav Dubeyko <slava@dubeyko.com> Cc: <stable@vger.kernel.org> Signed-off-by:
-
Sergei Antonov authored
Fix a bug which is reproduced as follows. Create a file: echo abc > test_file Try to expand the file beyond available space: truncate --size=<size exceeding available space> test_file Since HFS+ does not support file size > allocated size, truncate should fail. However, it ends successfully. The driver returns success despite having been unable to allocate the requested space for the file. Also filesystem check finds an error: Checking catalog file. Incorrect size for file test_file (It should be 469094400 instead of 1000000000) Add a piece of code analogous to code in the fat driver. Now a proper error is returned and filesystem remains consistent. Signed-off-by:
Sergei Antonov <saproj@gmail.com> Cc: Vyacheslav Dubeyko <slava@dubeyko.com> Cc: Hin-Tak Leung <htl10@users.sourceforge.net> Reviewed-by:
Anton Altaparmakov <anton@tuxera.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Christoph Hellwig <hch@infradead.org> Cc: Sougata Santra <sougata@tuxera.com> Signed-off-by:
-
Chengyu Song authored
In case of memory allocation error, the return should be -ENOMEM, instead of -ENOSPC. Signed-off-by:
Chengyu Song <csong84@gatech.edu> Reviewed-by:
-
Fabian Frederick authored
Signed-off-by:
Fabian Frederick <fabf@skynet.be> Signed-off-by:
-
Fabian Frederick authored
is_known_namespace() only returns true/false. Also remove inline and let compiler decide what to do with static functions. Signed-off-by:
-
Fabian Frederick authored
According to commit 5f16f322 ("ext4: atomically set inode->i_flags in ext4_set_inode_flags()"). Signed-off-by:
-
Fabian Frederick authored
security/trusted/user/osx setxattr did the same xattr_name initialization. Move that operation in hfsplus_setxattr(). Tested with security/trusted/user getfattr/setfattr Signed-off-by:
-
Fabian Frederick authored
security/trusted/user/osx getxattr did the same xattr_name initialization. Move that operation in hfsplus_getxattr(). Tested with security/trusted/user getfattr/setfattr Signed-off-by:
-
Dan Carpenter authored
This doesn't change how the code works, but clearly the curly braces were intended. Signed-off-by:
Dan Carpenter <dan.carpenter@oracle.com> Cc: Vyacheslav Dubeyko <slava@dubeyko.com> Cc: Sougata Santra <sougata@tuxera.com> Cc: Christoph Hellwig <hch@lst.de> Signed-off-by:
-
Chengyu Song authored
In case of memory allocation error, the return should be -ENOMEM, instead of -ENOSPC. Signed-off-by:
-
Ryusuke Konishi authored
Use inode_set_flags() to atomically set i_flags instead of clearing out the S_IMMUTABLE, S_APPEND, etc. flags and then setting them from the FS_IMMUTABLE_FL, FS_APPEND_FL flags to avoid a race where an immutable file has the immutable flag cleared for a brief window of time. This is a similar fix to commit 5f16f322 ("ext4: atomically set inode->i_flags in ext4_set_inode_flags()"). Signed-off-by:
Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp> Cc: "Theodore Ts'o" <tytso@mit.edu> Signed-off-by:
-
Ryusuke Konishi authored
nilfs_set_inode_flags() function adjusts gfp-mask of inode->i_mapping as well as i_flags, however, this coupling of operations is not appropriate. For instance, nilfs_ioctl_setflags(), one of three callers of nilfs_set_inode_flags(), doesn't need to reinitialize the gfp-mask at all. In addition, nilfs_new_inode(), another caller of nilfs_set_inode_flags(), doesn't either because it has already initialized the gfp-mask. Only __nilfs_read_inode(), the remaining caller, needs it. So, this moves the gfp mask manipulation to __nilfs_read_inode() from nilfs_set_inode_flags(). Signed-off-by:
-
Ryusuke Konishi authored
Fix the following build warning: fs/nilfs2/super.c: In function 'nilfs_checkpoint_is_mounted': fs/nilfs2/super.c:1023:10: warning: comparison of unsigned expression < 0 is always false [-Wtype-limits] if (cno < 0 || cno > nilfs->ns_cno) ^ This warning indicates that the comparision "cno < 0" is useless because variable "cno" has an unsigned integer type "__u64". Signed-off-by:David Binderman <dcb314@hotmail.com> Signed-off-by:
-
Ryusuke Konishi authored
The older a filesystem gets, the slower lscp command becomes. This is because nilfs_cpfile_do_get_cpinfo() function meets more hole blocks as the start offset of valid checkpoint numbers gets bigger. This reduces the overhead by skipping hole blocks efficiently with nilfs_mdt_find_block() helper. A measurement result of this patch is as follows: Before: $ time lscp CNO DATE TIME MODE FLG BLKCNT ICNT 5769303 2015-02-22 19:31:33 cp - 108 1 5769304 2015-02-22 19:38:54 cp - 108 1 real 0m0.182s user 0m0.003s sys 0m0.180s After: $ time lscp CNO DATE TIME MODE FLG BLKCNT ICNT 5769303 2015-02-22 19:31:33 cp - 108 1 5769304 2015-02-22 19:38:54 cp - 108 1 real 0m0.003s user 0m0.001s sys 0m0.002s Signed-off-by: -
Ryusuke Konishi authored
Add a new metadata file function, nilfs_mdt_find_block(), which finds an existent block on a metadata file in a given range of blocks. This function skips continuous hole blocks efficiently by using nilfs_bmap_seek_key(). Signed-off-by:
-
Ryusuke Konishi authored
Add a new bmap function, nilfs_bmap_seek_key(), which seeks a valid entry and returns its key starting from a given key. This function can be used to skip hole blocks efficiently. Signed-off-by:
-
Ryusuke Konishi authored
The type of key arguments in block mapping interface varies depending on function. For instance, nilfs_bmap_lookup_at_level() takes "__u64" for its key argument whereas nilfs_bmap_lookup() takes "unsigned long". This fits them to "__u64" to eliminate the variation. Signed-off-by:
-
Ryusuke Konishi authored
Simplify nilfs_mdt_bgl_lock() by utilizing bgl_lock_ptr() helper in <linux/blockgroup_lock.h>. Signed-off-by:
-
Ryusuke Konishi authored
nilfs_forget_buffer(), nilfs_clear_dirty_page(), and nilfs_segctor_complete_write() are using a bunch of atomic bit operations against buffer state bitmap. This reduces the number of them by utilizing set_mask_bits() macro. Signed-off-by:
-
Ryusuke Konishi authored
The async write flag is introduced to nilfs2 in the commit 7f42ec39 ("nilfs2: fix issue with race condition of competition between segments for dirty blocks"), but the flag only makes sense for data buffers and btree node buffers. It is not needed for segment summary buffers. This gets rid of the latter uses as part of refactoring of atomic bit operations on buffer state bitmap. Signed-off-by:
-
Fabian Frederick authored
See Documentation/CodingStyle Signed-off-by:
-
Fabian Frederick authored
See Documenation/CodingStyle Signed-off-by:
-
