- 17 Oct, 2007 40 commits
-
-
Oleg Nesterov authored
Two threads, T1 and T2. T2 ptraces P, and P is not a child of ptracer's thread group. P exits and goes to TASK_ZOMBIE. T1 does wait_task_zombie(P): P->exit_state = TASK_DEAD; ... read_unlock(&tasklist_lock); T2 does exit(), takes tasklist, forget_original_parent() does __ptrace_unlink(P) but doesn't call do_notify_parent(P) because p->exit_state == EXIT_DEAD. Now, P is not visible to our process: __ptrace_unlink() removed it from ->children. We should send notification to P->parent and release P if and only if SIGCHLD is ignored. And we have 3 bugs: 1. P->parent does do_wait() and gets -ECHILD (P is on ->parent->children, but its state is TASK_DEAD). 2. // wait_task_zombie() continues if (put_user(...)) { // TODO: is this safe? p->exit_state = EXIT_ZOMBIE; return; } we return without notification/release, task_struct leaked. Solution: ignore -EFAULT and proceed. It is an application's bug if we can't fill infop/stat_addr (in case of VM_FAULT_OOM we have much more problems). 3. // wait_task_zombie() continues if (p->real_parent != p->parent) { // Not taken, it was untraced'ed ... } release_task(p); we released the task which we shouldn't. Solution: check ->real_parent != ->parent before, under tasklist_lock, but use ptrace_unlink() instead of __ptrace_unlink() to check ->ptrace. This patch hopefully solves 2 and 3, the 1st bug will be fixed later, we need some cleanups in forget_original_parent/reparent_thread. However, the first race is very unlikely and not critical, so I hope it makes sense to fix 1 and 2 for now. 4. Small cleanup: don't "restore" EXIT_ZOMBIE unless we know we are not going to realease the child. Signed-off-by:Oleg Nesterov <oleg@tv-sign.ru> Cc: Ingo Molnar <mingo@elte.hu> Cc: Roland McGrath <roland@redhat.com> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
Oleg Nesterov authored
A zombie must have a valid ->signal, we are going to release it and __exit_signal() starts with BUG_ON(!sig). Signed-off-by:
-
Jesper Juhl authored
There's a small and unlikely memory leak in drivers/pnp/pnpbios/proc.c::proc_read_escd(). It's inside a sanity check, so it probably won't trigger often (if at all), however it *is* a potential leak and it's easy to avoid, so let's just fix it :) Signed-off-by:
Jesper Juhl <jesper.juhl@gmail.com> Signed-off-by:
-
Roland McGrath authored
This makes powerpc64's compat code use the new linux/elfcore-compat.h, reducing some hand-copied duplication. Signed-off-by:
Roland McGrath <roland@redhat.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Andi Kleen <ak@suse.de> Cc: "Luck, Tony" <tony.luck@intel.com> Signed-off-by:
-
Roland McGrath authored
This adds the linux/elfcore-compat.h header file, which is the CONFIG_COMPAT analog of the linux/elfcore.h header. Each arch that needs to fake out fs/binfmt_elf.c for its compat code can use this header to replace the hand-copied definitions of the compat variants of struct elf_prstatus et al. Only the pr_reg field varies by arch, so asm/{compat,elf}.h must define compat_elf_gregset_t before linux/elfcore-compat.h can be used. It's a clean-up that every arch with compat core dumping code can benefit from. I only touched the ones I have handy to test at home. Doing the same for each other arch should be straightforward, and I'm happy to offer tips. Signed-off-by: -
Satyam Sharma authored
The current code skips the check to verify whether the filesystem was previously cleanly unmounted, if (flags & UFS_ST_MASK) == UFS_ST_44BSD or UFS_ST_OLD. This looks like an inadvertent bug that slipped in due to parantheses in the compound conditional to me, especially given that ufs_get_fs_state() handles the UFS_ST_44BSD case perfectly well. So, let's fix the compound condition appropriately. Signed-off-by:
Satyam Sharma <satyam@infradead.org> Cc: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by:
-
Christoph Hellwig authored
Move prototypes and in-core structures to fs/ufs/ similar to what most other filesystems already do. I made little modifications: move also ufs debug macros and mount options constants into fs/ufs/ufs.h, this stuff also private for ufs. Signed-off-by:
Christoph Hellwig <hch@lst.de> Signed-off-by:
Evgeniy Dushistov <dushistov@mail.ru> Cc: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by:
-
Alexey Dobriyan authored
Driver does proc_mkdir("ipmi", NULL); but remove_proc_entry(proc_ipmi_root->name, &proc_root); This is OK and working if only slightly inconsistent. Also changing proc_root to NULL will help OpenVZ which has multiple proc roots and, as we now know, requires matching parents in such cases. Signed-off-by:Alexey Dobriyan <adobriyan@sw.ru> Cc: Corey Minyard <minyard@acm.org> Signed-off-by:
-
Oleg Nesterov authored
With or without this patch, multi-threaded init's are not fully supported, but do_exit() is completely wrong. This becomes a real problem when we support pid namespaces. 1. do_exit() panics when the main thread of /sbin/init exits. It should not until the whole thread group exits. Move the code below, under the "if (group_dead)" check. Note: this means that forget_original_parent() can use an already dead child_reaper()'s task_struct. This is OK for /sbin/init because - do_wait() from alive sub-thread still can reap a zombie, we iterate over all sub-thread's ->children lists - do_notify_parent() will wakeup some alive sub-thread because it sends the group-wide signal However, we should remove choose_new_parent()->BUG_ON(reaper->exit_state) for this. 2. We are playing games with ->nsproxy->pid_ns. This code is bogus today, and it has to be changed anyway when we really support pid namespaces, just remove it. Signed-off-by:
-
Andi Kleen authored
The Linux ELF loader is quite complicated and messy code (that could probably need a rewrite, but that's a different chapter). One particular messy part in it is the support for non ELF a.out ld.sos. This was originally added to make transition from a.out to ELF easier because an a.out ELF ld.so could be still build using an older a.out toolkit. But by now that should be fully obsolete and removing it would clean up binfmt_elf.c up a bit. I propose to deprecate this support and remove for 2.6.25. Drawback is that someone still runs their system with a.out ld.so they would need to update the ld.so when updating to a new kernel. This patch just adds an entry to the deprecation file and a printk warning users. [akpm@linux-foundation.org: better warning message] Signed-off-by:
Andi Kleen <ak@suse.de> Signed-off-by:
-
Oleg Nesterov authored
With the recent changes, do_sigaction()->recalc_sigpending_and_wake() can never clear TIF_SIGPENDING. Instead, it can set this flag and wake up the thread without any reason. Harmless, but unneeded and wastes CPU. Signed-off-by:
-
Oleg Nesterov authored
It is a bit annoying that do_exit() takes ->pi_lock to set PF_EXITING. All we need is to synchronize with lookup_pi_state() which saw this task without PF_EXITING under ->pi_lock. Change do_exit() to use spin_unlock_wait(). Signed-off-by:
Ingo Molnar <mingo@elte.hu> Signed-off-by:
-
Bjorn Helgaas authored
If we discover the TIS TPM device via PNP, use the PNP IRQ information rather than probing for an IRQ. If PNP shows no IRQ, run the TPM in polling mode. Tested-by: <valdis.kletnieks@vt.edu> Signed-off-by:
Bjorn Helgaas <bjorn.helgaas@hp.com> Cc: Kylene Hall <kjhall@us.ibm.com> Cc: <tpm@selhorst.net> Signed-off-by:
-
Mariusz Kozlowski authored
drivers/char/ip2/ip2main.c | 104398 -> 104346 (-52 bytes) drivers/char/ip2/ip2main.o | 210710 -> 210702 (-8 bytes) Signed-off-by:
Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by:
-
Mariusz Kozlowski authored
fs/autofs4/inode.c | 10467 -> 10435 (-32 bytes) fs/autofs4/inode.o | 98576 -> 98552 (-24 bytes) Signed-off-by:
Ian Kent <raven@themaw.net> Signed-off-by:
-
Mariusz Kozlowski authored
Documentation/firmware_class/firmware_sample_firmware_class.c | 5246 -> 5211 (-35 bytes) Signed-off-by:
-
Mariusz Kozlowski authored
drivers/char/consolemap.c | 22678 -> 22650 (-28 bytes) drivers/char/consolemap.o | 90113 -> 90029 (-84 bytes) Signed-off-by:
Alan Cox <alan@redhat.com> Signed-off-by:
-
Mike Frysinger authored
Add two new functions for reading the kernel log buffer. The intention is for them to be used by recovery/dump/debug code so the kernel log can be easily retrieved/parsed in a crash scenario, but they are generic enough for other people to dream up other fun uses. [akpm@linux-foundation.org: buncha fixes] Signed-off-by:
Mike Frysinger <vapier@gentoo.org> Cc: Robin Getz <rgetz@blackfin.uclinux.org> Cc: Greg Ungerer <gerg@snapgear.com> Cc: Russell King <rmk@arm.linux.org.uk> Cc: Paul Mundt <lethal@linux-sh.org> Acked-by:
Tim Bird <tim.bird@am.sony.com> Signed-off-by:
-
Adrian Bunk authored
This patch makes two needlessly global functions static. Signed-off-by:
Adrian Bunk <bunk@stusta.de> Signed-off-by:
-
Adrian Bunk authored
Every file should include the headers containing the prototypes for its global functions (in this case ioremap_page_range()). Signed-off-by:
Haavard Skinnemoen <hskinnemoen@atmel.com> Signed-off-by:
-
Adrian Bunk authored
This patch contains the following possible cleanups: - make the following needlessly global functions static: - rxrpc.c: afs_send_pages() - vlocation.c: afs_vlocation_queue_for_updates() - write.c: afs_writepages_region() - make the following needlessly global variables static: - mntpt.c: afs_mntpt_expiry_timeout - proc.c: afs_vlocation_states[] - server.c: afs_server_timeout - vlocation.c: afs_vlocation_timeout - vlocation.c: afs_vlocation_update_timeout - #if 0 the following unused function: - cell.c: afs_get_cell_maybe() - #if 0 the following unused variables: - callback.c: afs_vnode_update_timeout - cmservice.c: struct afs_cm_workqueue Signed-off-by:
David Howells <dhowells@redhat.com> Signed-off-by:
-
Adrian Bunk authored
This patch contains the following cleanups: - make the needlessly global variable rt_trace_on static - remove the unused global function deadlock_trace_off() Signed-off-by:
-
Roland McGrath authored
This patch adds the /sys/module/<name>/notes/ magic directory, which has a file for each allocated SHT_NOTE section that appears in <name>.ko. This is the counterpart for each module of /sys/kernel/notes for vmlinux. Reading this delivers the contents of the module's SHT_NOTE sections. This lets userland easily glean any detailed information about that module's build that was stored there at compile time (e.g. by ld --build-id). Signed-off-by:
-
Alexey Dobriyan authored
* Remove stupid comments, like, at the beginning of every function that function begins (twice per function) and at the end (once) * Remove trailing or otherwise broken whitespace as per let c_space_errors=1 * Reformat comments to fit it into 80 columns and remove stupid ------------'s. * Indent case labels on the same column where switch begins * other minor CS tweaks not worth mentioning Signed-off-by:
Alexey Dobriyan <adobriyan@gmail.com> Cc: Jiri Slaby <jirislaby@gmail.com> Signed-off-by:
-
Neil Horman authored
Fix do_coredump to detect a crash in the user mode helper process and abort the attempt to recursively dump core to another copy of the helper process, potentially ad-infinitum. [akpm@linux-foundation.org: cleanups] Signed-off-by:
Neil Horman <nhorman@tuxdriver.com> Cc: <martin.pitt@ubuntu.com> Cc: <wwoods@redhat.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Signed-off-by:
-
Neil Horman authored
A rewrite of my previous post for this enhancement. It uses jeremy's split_argv/free_argv library functions to translate core_pattern into an argv array to be passed to the user mode helper process. It also adds a translation to format_corename such that the origional value of RLIMIT_CORE can be passed to userspace as an argument. Signed-off-by:
-
Neil Horman authored
For some time /proc/sys/kernel/core_pattern has been able to set its output destination as a pipe, allowing a user space helper to receive and intellegently process a core. This infrastructure however has some shortcommings which can be enhanced. Specifically: 1) The coredump code in the kernel should ignore RLIMIT_CORE limitation when core_pattern is a pipe, since file system resources are not being consumed in this case, unless the user application wishes to save the core, at which point the app is restricted by usual file system limits and restrictions. 2) The core_pattern code should be able to parse and pass options to the user space helper as an argv array. The real core limit of the uid of the crashing proces should also be passable to the user space helper (since it is overridden to zero when called). 3) Some miscellaneous bugs need to be cleaned up (specifically the recognition of a recursive core dump, should the user mode helper itself crash. Also, the core dump code in the kernel should not wait for the user mode helper to exit, since the same context is responsible for writing to the pipe, and a read of the pipe by the user mode helper will result in a deadlock. This patch: Remove the check of RLIMIT_CORE if core_pattern is a pipe. In the event that core_pattern is a pipe, the entire core will be fed to the user mode helper. Signed-off-by:
-
Neil Horman authored
It would be nice if the argv_split library function could gracefully handle a NULL pointer in the argcp parameter, so as to allow functions using it that did not care about the value of argc to not have to declare a useless variable. This patch accomplishes that. Tested by me, with successful results. Signed-off-by:
Jeremy Fitzhardinge <jeremy@goop.org> Cc: Satyam Sharma <satyam@infradead.org> Signed-off-by:
-
Evgeniy Dushistov authored
An implementation of show_options method for UFS. Signed-off-by:
-
Mark Fortescue authored
Add in support for SunOS 4.1.x flavor of BSD 4.2 UFS filing system Macros have been put in to alow suport for the old static table Cylinder Groups but this implementation does not use them yet. This also fixes Solaris UFS filing system access by disabling fast symbolic links as Sun's version of UFS does not support on-disk fast symbolic links. Tested by: Ppartitioning a new disk using SunOS 4.1.1, creating a UFS filing system on one of the partitions and writing some files to the filing system. Using Linux-2.6.22 (patched) to read the files and then write a shed load of files to the UFS partition. Using SunOS 4.1.1 to verify the filing system is OK and to check the files. The test host is a sun4c SS1 Clone. [akpm@linux-foundation.org: coding style fixes] [adobriyan@gmail.com: fix oops] Signed-off-by:
Mark Fortescue <mark@mtfhpc.demon.co.uk> Cc: Evgeniy Dushistov <dushistov@mail.ru> Signed-off-by:
-
Eric Sandeen authored
ext[234]_get_group_desc never tests the bh argument, and only sets it if it is passed in; it is perfectly happy with a NULL bh argument. But, many callers send one in and never use it. May as well call with NULL like other callers who don't use the bh. Signed-off-by:
Eric Sandeen <sandeen@redhat.com> Cc: <linux-ext4@vger.kernel.org> Signed-off-by:
-
Denis Cheng authored
Since the mempages parameter is actually not used, they should be removed. Now there is only files_init use the mempages parameter, files_init(mempages); but I don't think the adaptation to mempages in files_init is really useful; and if files_init also changed to the prototype void (*func)(void), the wrapper vfs_caches_init would also not need the mempages parameter. Signed-off-by:
Denis Cheng <crquan@gmail.com> Signed-off-by:
-
David Woodhouse authored
Andy Gospodarek pointed out that because we return in the middle of the free_irq() function, we never actually do call the IRQ handler that just got deregistered. This should fix it, although I expect Andrew will want to convert those 'return's to 'break'. That's a separate change though. Signed-off-by:
David Woodhouse <dwmw2@infradead.org> Cc: Andy Gospodarek <andy@greyhouse.net> Cc: Fernando Luis Vzquez Cao <fernando@oss.ntt.co.jp> Signed-off-by:
-
Ravikiran G Thirumalai authored
Commit 4004c69a avoids too many remote cpu references while reporting per-irq stats. Since we will not have the same performance penalty of bringing in remote cpu cachelines while reporting per-irq stats anymore, we can now afford to be consistent and report this statistic on all arches, all configs. akpm: affects ia64, alpha and ppc64, mainly. Kiran earlier said: Read to /proc/stat takes: Plain: 2.622832 With speedup patch: 0.013194 With the per-irq stats commented out: 0.008124 So the performance problems which originally caused those architectures to disable this statistic should now be fixed up. Signed-off-by:
Ravikiran Thirumalai <kiran@scalex86.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: "Luck, Tony" <tony.luck@intel.com> Cc: Richard Henderson <rth@twiddle.net> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Acked-by:
-
Rusty Russell authored
Adrian Bunk points out that "unsafe" was used to mark modules touched by the deprecated MOD_INC_USE_COUNT interface, which has long gone. It's time to remove the member from the module structure, as well. If you want a module which can't unload, don't register an exit function. (Vlad Yasevich says SCTP is now safe to unload, so just remove the __unsafe there). Signed-off-by:
Rusty Russell <rusty@rustcorp.com.au> Acked-by:
Shannon Nelson <shannon.nelson@intel.com> Acked-by:
Dan Williams <dan.j.williams@intel.com> Acked-by:
Vlad Yasevich <vladislav.yasevich@hp.com> Cc: Sridhar Samudrala <sri@us.ibm.com> Cc: Adrian Bunk <bunk@stusta.de> Signed-off-by:
-
Miklos Szeredi authored
Signed-off-by:
Miklos Szeredi <mszeredi@suse.cz> Cc: <linux-ext4@vger.kernel.org> Signed-off-by:
-
Miklos Szeredi authored
Signed-off-by:
-
Miklos Szeredi authored
Using mtab is problematic for various reasons, one of them is that unprivileged mounts won't turn up in there. So we want to get rid of it, and use /proc/mounts instead. But most filesystems are lazy, and are not showing all mount options. Which means, that without mtab, the user won't be able to see some or all of the options. It would be nice if the generic code could remember the mount options, and show them without the need to add extra code to filesystems. But this is not easy, because different filesystems handle mount options given options, and not tough the rest. This is not taken into account by mount(8) either, so /etc/mtab will be broken in this case. This series fixes up ->show_options() in ext[234]. Signed-off-by:
-
Fengguang Wu authored
It's *wrong* to have #define log2(n) ffz(~(n)) It should be *reversed*: #define log2(n) flz(~(n)) or #define log2(n) fls(n) or just use ilog2(n) defined in linux/log2.h. This patch follows the last solution, recommended by Andrew Morton. Cc: <linux-ext4@vger.kernel.org> Cc: Mingming Cao <cmm@us.ibm.com> Cc: Bjorn Helgaas <bjorn.helgaas@hp.com> Cc: Chris Ahna <christopher.j.ahna@intel.com> Cc: David Mosberger-Tang <davidm@hpl.hp.com> Cc: Kyle McMartin <kyle@parisc-linux.org> Cc: Dave Airlie <airlied@linux.ie> Cc: Dave Jones <davej@codemonkey.org.uk> Signed-off-by:
Fengguang Wu <wfg@mail.ustc.edu.cn> Signed-off-by:
-
Avi Kivity authored
smp_call_function_single() now knows how to call the function on the current cpu. Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ingo Molnar <mingo@elte.hu> Signed-off-by:
Avi Kivity <avi@qumranet.com> Signed-off-by:
-
