- 23 Sep, 2009 40 commits
-
-
Adrian Hunter authored
Signed-off-by:
Adrian Hunter <adrian.hunter@nokia.com> Acked-by:
Matt Fleming <matt@console-pimps.org> Cc: Ian Molton <ian@mnementh.co.uk> Cc: "Roberto A. Foglietta" <roberto.foglietta@gmail.com> Cc: Jarkko Lavinen <jarkko.lavinen@nokia.com> Cc: Denis Karpov <ext-denis.2.karpov@nokia.com> Cc: Pierre Ossman <pierre@ossman.eu> Cc: Philip Langdale <philipl@overt.org> Cc: "Madhusudhan" <madhu.cr@ti.com> Cc: <linux-mmc@vger.kernel.org> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
Denis Karpov authored
The code could set the bit to 1 but not reset it to 0. Signed-off-by:
Denis Karpov <ext-denis.2.karpov@nokia.com> Signed-off-by:
-
Denis Karpov authored
Keep the context over PM dynamic OFF states. Signed-off-by:
-
Adrian Hunter authored
This patch is preparation for adding context save and restore support. Keep track of the current power mode so that the context restore function can avoid restoring the context for a card if the power has been switched off. If the power is off, the card must be reinitialized anyway which will re-establish the context. Signed-off-by:
-
Denis Karpov authored
PM dynamic OFF state results in context loss. That is, the host controller has been powered off at some point, which means the registers have been reset. The driver must detect when this happens, and restore the context. This patch adds the means to detect context loss. Note, the PM side is not yet implemented. Signed-off-by:
-
Adrian Hunter authored
For the moment enable / disable just turns the fclk on and off. Signed-off-by:
-
Denis Karpov authored
Adds <debugfs_root>/kernel/debug/mmc<N>/regs entry, contents show registers' state and some driver internal state variables. Signed-off-by:
-
Adrian Hunter authored
According to the standard, the SWITCH command should be followed by a SEND_STATUS command to check for errors. Signed-off-by:
-
Denis Karpov authored
Fix MMC host stop sequence: power off once. Signed-off-by:
-
Jarkko Lavinen authored
Add support for the new MMC command SLEEP_AWAKE. Signed-off-by:
Jarkko Lavinen <jarkko.lavinen@nokia.com> Signed-off-by:
-
Adrian Hunter authored
Power can be saved by powering off cards that are not in use. This is similar to suspend / resume except it is under the control of the driver, and does not require any power management support. It can only be used when the driver can monitor whether the card is removed, otherwise it is unsafe. This is possible because, unlike suspend, the driver still receives card detect and / or cover switch interrupts. Signed-off-by:
-
Adrian Hunter authored
eMMC's are not removable, so unsafe resume is OK always. To permit this a new host capability MMC_CAP_NONREMOVABLE has been added and suspend / resume updated accordingly. Signed-off-by:
-
Adrian Hunter authored
This change allows the MMC host to be claimed in situations where the host may or may not have already been claimed. Also 'mmc_try_claim_host()' is now exported. Signed-off-by:
-
Adrian Hunter authored
MMC hosts that support power saving can use the 'enable' and 'disable' methods to exit and enter power saving states. An explanation of their use is provided in the comments added to include/linux/mmc/host.h. Signed-off-by:
-
Ohad Ben-Cohen authored
This is needed for 1.8V embedded SDIO devices and supporting host controllers (e.g. TI 127x and ZOOM2 boards) Signed-off-by:
Ohad Ben-Cohen <ohad@bencohen.org> Acked-by:
-
Uwe Kleine-König authored
omap_mmc_probe lives in .init.text, so using platform_driver_register to register it is wrong because binding a device after the init memory is discarded (e.g. via sysfs) results in an oops. As requested by David Brownell platform_driver_probe is used instead of moving the probe function to .devinit.text as proposed initially. This saves some memory, but devices registered after the driver is probed are not bound (probably there are none) and binding via sysfs isn't possible. Signed-off-by:
Uwe Kleine-König <u.kleine-koenig@pengutronix.de> Cc: Jean Pihet <jpihet@mvista.com> Cc: Tony Lindgren <tony@atomide.com> Cc: Pierre Ossman <drzeus@drzeus.cx> Cc: Andy Lowe <alowe@mvista.com> Cc: Adrian Hunter <ext-adrian.hunter@nokia.com> Cc: Andrew Morton <akpm@linux-foundation.org> Acked-by:
David Brownell <dbrownell@users.sourceforge.net> Cc: Madhusudhan Chikkature<madhu.cr@ti.com> Cc: Greg Kroah-Hartman <gregkh@suse.de> Cc: Matt Fleming <matt@console-pimps.org> Cc: Ian Molton <ian@mnementh.co.uk> Cc: "Roberto A. Foglietta" <roberto.foglietta@gmail.com> Cc: Philip Langdale <philipl@overt.org> Signed-off-by:
-
Joe Perches authored
Signed-off-by:
Joe Perches <joe@perches.com> Cc: Pavel Machek <pavel@ucw.cz> Cc: Brian Swetland <swetland@google.com> Cc: Pierre Ossman <drzeus-list@drzeus.cx> Cc: San Mehat <san@android.com> Cc: Matt Fleming <matt@console-pimps.org> Cc: Ian Molton <ian@mnementh.co.uk> Cc: "Roberto A. Foglietta" <roberto.foglietta@gmail.com> Cc: Philip Langdale <philipl@overt.org> Signed-off-by:
-
Joe Perches authored
Make it a bit more like typical kernel style. Signed-off-by:
-
Joe Perches authored
Signed-off-by:
-
San Mehat authored
MMC Driver for HTC Dream. I picked the code up from Google git trees, removed stuff not strictly necessary, and did a few cleanups. It still works :-). Signed-off-by:
Pavel Machek <pavel@ucw.cz> Cc: Brian Swetland <swetland@google.com> Cc: Pierre Ossman <drzeus-list@drzeus.cx> Cc: Joe Perches <joe@perches.com> Cc: Matt Fleming <matt@console-pimps.org> Cc: Ian Molton <ian@mnementh.co.uk> Cc: "Roberto A. Foglietta" <roberto.foglietta@gmail.com> Cc: Philip Langdale <philipl@overt.org> Signed-off-by:
-
Anand Gadiyar authored
This considerably reduces the number of interrupts during a transfer and ought to result in some power saving. Signed-off-by:
Anand Gadiyar <gadiyar@ti.com> Signed-off-by:
Santosh Shilimkar <santosh.shilimkar@ti.com> Acked-by:
Kishore Kadiyala <kishore.kadiyala@ti.com> Cc: Pierre Ossman <drzeus@drzeus.cx> Cc: Matt Fleming <matt@console-pimps.org> Cc: Ian Molton <ian@mnementh.co.uk> Cc: "Roberto A. Foglietta" <roberto.foglietta@gmail.com> Cc: Philip Langdale <philipl@overt.org> Signed-off-by:
-
Balaji Rao authored
When mmc_power_up is called during unsafe resume, host->ocr should be used instead of host->ocr_avail. Signed-off-by:
Balaji Rao <balajirrao@openmoko.org> Cc: Andy Green <andy@openmoko.com> Cc: Pierre Ossman <drzeus-mmc@drzeus.cx> Cc: Ian Molton <ian@mnementh.co.uk> Cc: "Roberto A. Foglietta" <roberto.foglietta@gmail.com> Cc: Philip Langdale <philipl@overt.org> Acked-by:
-
Mike Frysinger authored
Signed-off-by:
Mike Frysinger <vapier@gentoo.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Robin Getz <rgetz@blackfin.uclinux.org> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Arnd Bergmann <arnd@arndb.de> Signed-off-by:
-
Mike Frysinger authored
This allows lockdep to locate symbols that are in arch-specific data sections (such as data in Blackfin on-chip SRAM regions). Signed-off-by:
-
Mike Frysinger authored
This allows kallsyms to locate symbols that are in arch-specific text sections (such as text in Blackfin on-chip SRAM regions). Signed-off-by:
-
Mike Frysinger authored
Some ports (like the Blackfin arch) have a discontiguous memory map which means there may be text or data that falls outside of the standard range of the start/end text/data symbols. Creating some helper functions allows these non-standard ports to declare these regions without adversely affecting anyone else. Signed-off-by:
-
Paul Mundt authored
Commit b478b782 "kallsyms, tracing: output more proper symbol name" introduces a "bugfix" that introduces a segfault in kallsyms in my configurations. The cause is the introduction of prefix_underscores_count() which attempts to count underscores, even in symbols that do not have them. As a result, it just uselessly runs past the end of the buffer until it crashes: CC init/version.o LD init/built-in.o LD .tmp_vmlinux1 KSYM .tmp_kallsyms1.S /bin/sh: line 1: 16934 Done sh-linux-gnu-nm -n .tmp_vmlinux1 16935 Segmentation fault | scripts/kallsyms > .tmp_kallsyms1.S make: *** [.tmp_kallsyms1.S] Error 139 This simplifies the logic and just does a straightforward count. Signed-off-by:
Paul Mundt <lethal@linux-sh.org> Reviewed-by:
Li Zefan <lizf@cn.fujitsu.com> Cc: Lai Jiangshan <laijs@cn.fujitsu.com> Cc: Sam Ravnborg <sam@ravnborg.org> Cc: Paulo Marques <pmarques@grupopie.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: <stable@kernel.org> [2.6.30.x, 2.6.31.x] Signed-off-by:
-
Jiri Pirko authored
Make ->ru_maxrss value in struct rusage filled accordingly to rss hiwater mark. This struct is filled as a parameter to getrusage syscall. ->ru_maxrss value is set to KBs which is the way it is done in BSD systems. /usr/bin/time (gnu time) application converts ->ru_maxrss to KBs which seems to be incorrect behavior. Maintainer of this util was notified by me with the patch which corrects it and cc'ed. To make this happen we extend struct signal_struct by two fields. The first one is ->maxrss which we use to store rss hiwater of the task. The second one is ->cmaxrss which we use to store highest rss hiwater of all task childs. These values are used in k_getrusage() to actually fill ->ru_maxrss. k_getrusage() uses current rss hiwater value directly if mm struct exists. Note: exec() clear mm->hiwater_rss, but doesn't clear sig->maxrss. it is intetionally behavior. *BSD getrusage have exec() inheriting. test programs ======================================================== getrusage.c =========== #include <stdio.h> #include <stdlib.h> #include <string.h> #include <sys/types.h> #include <sys/time.h> #include <sys/resource.h> #include <sys/types.h> #include <sys/wait.h> #include <unistd.h> #include <signal.h> #include <sys/mman.h> #include "common.h" #define err(str) perror(str), exit(1) int main(int argc, char** argv) { int status; printf("allocate 100MB\n"); consume(100); printf("testcase1: fork inherit? \n"); printf(" expect: initial.self ~= child.self\n"); show_rusage("initial"); if (__fork()) { wait(&status); } else { show_rusage("fork child"); _exit(0); } printf("\n"); printf("testcase2: fork inherit? (cont.) \n"); printf(" expect: initial.children ~= 100MB, but child.children = 0\n"); show_rusage("initial"); if (__fork()) { wait(&status); } else { show_rusage("child"); _exit(0); } printf("\n"); printf("testcase3: fork + malloc \n"); printf(" expect: child.self ~= initial.self + 50MB\n"); show_rusage("initial"); if (__fork()) { wait(&status); } else { printf("allocate +50MB\n"); consume(50); show_rusage("fork child"); _exit(0); } printf("\n"); printf("testcase4: grandchild maxrss\n"); printf(" expect: post_wait.children ~= 300MB\n"); show_rusage("initial"); if (__fork()) { wait(&status); show_rusage("post_wait"); } else { system("./child -n 0 -g 300"); _exit(0); } printf("\n"); printf("testcase5: zombie\n"); printf(" expect: pre_wait ~= initial, IOW the zombie process is not accounted.\n"); printf(" post_wait ~= 400MB, IOW wait() collect child's max_rss. \n"); show_rusage("initial"); if (__fork()) { sleep(1); /* children become zombie */ show_rusage("pre_wait"); wait(&status); show_rusage("post_wait"); } else { system("./child -n 400"); _exit(0); } printf("\n"); printf("testcase6: SIG_IGN\n"); printf(" expect: initial ~= after_zombie (child's 500MB alloc should be ignored).\n"); show_rusage("initial"); signal(SIGCHLD, SIG_IGN); if (__fork()) { sleep(1); /* children become zombie */ show_rusage("after_zombie"); } else { system("./child -n 500"); _exit(0); } printf("\n"); signal(SIGCHLD, SIG_DFL); printf("testcase7: exec (without fork) \n"); printf(" expect: initial ~= exec \n"); show_rusage("initial"); execl("./child", "child", "-v", NULL); return 0; } child.c ======= #include <sys/types.h> #include <unistd.h> #include <sys/types.h> #include <sys/wait.h> #include <stdio.h> #include <stdlib.h> #include <string.h> #include <sys/types.h> #include <sys/time.h> #include <sys/resource.h> #include "common.h" int main(int argc, char** argv) { int status; int c; long consume_size = 0; long grandchild_consume_size = 0; int show = 0; while ((c = getopt(argc, argv, "n:g:v")) != -1) { switch (c) { case 'n': consume_size = atol(optarg); break; case 'v': show = 1; break; case 'g': grandchild_consume_size = atol(optarg); break; default: break; } } if (show) show_rusage("exec"); if (consume_size) { printf("child alloc %ldMB\n", consume_size); consume(consume_size); } if (grandchild_consume_size) { if (fork()) { wait(&status); } else { printf("grandchild alloc %ldMB\n", grandchild_consume_size); consume(grandchild_consume_size); exit(0); } } return 0; } common.c ======== #include <stdio.h> #include <stdlib.h> #include <string.h> #include <sys/types.h> #include <sys/time.h> #include <sys/resource.h> #include <sys/types.h> #include <sys/wait.h> #include <unistd.h> #include <signal.h> #include <sys/mman.h> #include "common.h" #define err(str) perror(str), exit(1) void show_rusage(char *prefix) { int err, err2; struct rusage rusage_self; struct rusage rusage_children; printf("%s: ", prefix); err = getrusage(RUSAGE_SELF, &rusage_self); if (!err) printf("self %ld ", rusage_self.ru_maxrss); err2 = getrusage(RUSAGE_CHILDREN, &rusage_children); if (!err2) printf("children %ld ", rusage_children.ru_maxrss); printf("\n"); } /* Some buggy OS need this worthless CPU waste. */ void make_pagefault(void) { void *addr; int size = getpagesize(); int i; for (i=0; i<1000; i++) { addr = mmap(NULL, size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANON, -1, 0); if (addr == MAP_FAILED) err("make_pagefault"); memset(addr, 0, size); munmap(addr, size); } } void consume(int mega) { size_t sz = mega * 1024 * 1024; void *ptr; ptr = malloc(sz); memset(ptr, 0, sz); make_pagefault(); } pid_t __fork(void) { pid_t pid; pid = fork(); make_pagefault(); return pid; } common.h ======== void show_rusage(char *prefix); void make_pagefault(void); void consume(int mega); pid_t __fork(void); FreeBSD result (expected result) ======================================================== allocate 100MB testcase1: fork inherit? expect: initial.self ~= child.self initial: self 103492 children 0 fork child: self 103540 children 0 testcase2: fork inherit? (cont.) expect: initial.children ~= 100MB, but child.children = 0 initial: self 103540 children 103540 child: self 103564 children 0 testcase3: fork + malloc expect: child.self ~= initial.self + 50MB initial: self 103564 children 103564 allocate +50MB fork child: self 154860 children 0 testcase4: grandchild maxrss expect: post_wait.children ~= 300MB initial: self 103564 children 154860 grandchild alloc 300MB post_wait: self 103564 children 308720 testcase5: zombie expect: pre_wait ~= initial, IOW the zombie process is not accounted. post_wait ~= 400MB, IOW wait() collect child's max_rss. initial: self 103564 children 308720 child alloc 400MB pre_wait: self 103564 children 308720 post_wait: self 103564 children 411312 testcase6: SIG_IGN expect: initial ~= after_zombie (child's 500MB alloc should be ignored). initial: self 103564 children 411312 child alloc 500MB after_zombie: self 103624 children 411312 testcase7: exec (without fork) expect: initial ~= exec initial: self 103624 children 411312 exec: self 103624 children 411312 Linux result (actual test result) ======================================================== allocate 100MB testcase1: fork inherit? expect: initial.self ~= child.self initial: self 102848 children 0 fork child: self 102572 children 0 testcase2: fork inherit? (cont.) expect: initial.children ~= 100MB, but child.children = 0 initial: self 102876 children 102644 child: self 102572 children 0 testcase3: fork + malloc expect: child.self ~= initial.self + 50MB initial: self 102876 children 102644 allocate +50MB fork child: self 153804 children 0 testcase4: grandchild maxrss expect: post_wait.children ~= 300MB initial: self 102876 children 153864 grandchild alloc 300MB post_wait: self 102876 children 307536 testcase5: zombie expect: pre_wait ~= initial, IOW the zombie process is not accounted. post_wait ~= 400MB, IOW wait() collect child's max_rss. initial: self 102876 children 307536 child alloc 400MB pre_wait: self 102876 children 307536 post_wait: self 102876 children 410076 testcase6: SIG_IGN expect: initial ~= after_zombie (child's 500MB alloc should be ignored). initial: self 102876 children 410076 child alloc 500MB after_zombie: self 102880 children 410076 testcase7: exec (without fork) expect: initial ~= exec initial: self 102880 children 410076 exec: self 102880 children 410076 Signed-off-by:Jiri Pirko <jpirko@redhat.com> Signed-off-by:
KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Hugh Dickins <hugh.dickins@tiscali.co.uk> Cc: Ingo Molnar <mingo@elte.hu> Signed-off-by:
-
Andi Kleen authored
I tend to use a 'D' debugging macro a lot during debugging. When I define it before includes I often get conflicts with kmap_types.h's use of 'D' too. It's not very nice when a global include pollutes the name space like this. Rename the kmap_types.h D to KMAP_D. It is only used temporarily in the header so has no effect on anything else. Signed-off-by:
Andi Kleen <ak@linux.intel.com> Reviewed-by:
WANG Cong <xiyou.wangcong@gmail.com> Signed-off-by:
-
Rolf Eike Beer authored
abs() will truncate the input if is it outside the 2^32 range. Fix that by assuming `long' input. This might generate worse code in the common case. Signed-off-by:
Rolf Eike Beer <eike-kernel@sf-tec.de> Signed-off-by:
-
Suzuki Poulose authored
Compat utimensat() returns EINVAL when the tv_nsec is one of UTIME_OMIT or UTIME_NOW and the tv_sec is set to non-zero. As per man pages, the tv_sec field should be ignored. sys_utimensat() works fine in this case. Test case: #define _GNU_SOURCE #define _ATFILE_SOURCE #include <stdio.h> #include <fcntl.h> #include <unistd.h> #include <sys/stat.h> #include <stdlib.h> main(int argc, char *argv[]) { struct timespec ts[2]; struct timespec *tsp; if (argc < 2) { fprintf(stderr, "Usage : %s filename\n", argv[0]); exit (-1); } ts[0].tv_nsec = ts[1].tv_nsec = UTIME_NOW; ts[0].tv_sec = ts[1].tv_sec = 1; tsp = ts; if (utimensat(AT_FDCWD, argv[1],tsp,0) == -1) perror("utimensat"); else fprintf(stdout, "utimensat success\n"); return 0; } mjs22lp5:~ # cc -m64 utimensat-test.c -o utimensat_test64 mjs22lp5:~ # cc -m32 utimensat-test.c -o utimensat_test32 mjs22lp5:~ # ./utimensat_test32 /tmp/utimensat_test utimensat: Invalid argument mjs22lp5:~ # ./utimensat_test64 /tmp/utimensat_test utimensat success mjs22lp5:~ # uname -r 2.6.31-rc8 With the patch : mjs22lp5:~ # ./utimensat_test64 /tmp/utimensat_test utimensat success mjs22lp5:~ # ./utimensat_test32 /tmp/utimensat_test utimensat success mjs22lp5:~ # uname -r 2.6.31-rc8utimensat Signed-off-by:Suzuki K P <suzuki@in.ibm.com> Cc: Ulrich Drepper <drepper@redhat.com> Signed-off-by:
-
Jaswinder Singh Rajput authored
Fix the following 'make includecheck' warning: drivers/vlynq/vlynq.c: linux/device.h is included more than once. Signed-off-by:
Jaswinder Singh Rajput <jaswinderrajput@gmail.com> Signed-off-by:
Florian Fainelli <florian@openwrt.org> Signed-off-by:
-
Christoph Hellwig authored
qnx4 wrte support has never been fully implement, is broken since the dawn of time and hasn't been actively developed since before git history started. Instead of letting it further bitrot and complicate API transition (like the new truncate code) remove it. Signed-off-by:
Christoph Hellwig <hch@lst.de> Cc: Anders Larsen <al@alarsen.net> Cc: Nick Piggin <npiggin@suse.de> Signed-off-by:
-
Christoph Hellwig authored
do_sync_write() does the right thing for turning the aio_writev method into a normal non-vectored synchronous write, no need to duplicate it in ntfs. Signed-off-by:
Anton Altaparmakov <aia21@cantab.net> Signed-off-by:
-
Davide Libenzi authored
Split the anonfd interface into a bare file pointer creation one, and a file pointer creation plus install one. There are cases, like the usage of eventfds inside other kernel interfaces, where the file pointer created by anonfd needs to be used inside the initialization of other structures. As it is right now, as soon as anon_inode_getfd() returns, the kenrle can race with userspace closing the newly installed file descriptor. This patch, while keeping the old anon_inode_getfd(), introduces a new anon_inode_getfile() (whose services are reused in anon_inode_getfd()) that allows to split the file creation phase and the fd install one. Once all the kernel structures are initialized, the code can call the proper fd_install(). Gregory manifested the need for something like this inside KVM. Signed-off-by:
Davide Libenzi <davidel@xmailserver.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: James Morris <jmorris@namei.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Gregory Haskins <ghaskins@novell.com> Acked-by:
Serge Hallyn <serue@us.ibm.com> Acked-by:
Roland Dreier <rolandd@cisco.com> Signed-off-by:
-
Bartlomiej Zolnierkiewicz authored
On Saturday 01 August 2009 00:30:39 Mail Delivery Subsystem wrote: > Delivery to the following recipient failed permanently: > > linware@sh.cvut.cz > > Technical details of permanent failure: > Google tried to deliver your message, but it was rejected by the recipient > domain. We recommend contacting the other email provider for further > information about the cause of this error. The error that the other server > returned was: 450 450 <linware@sh.cvut.cz>: Recipient address rejected: > undeliverable address: unknown user: "linware" (state 14). Cc: Petr Vandrovec <vandrove@vc.cvut.cz> Signed-off-by:
Bartlomiej Zolnierkiewicz <bzolnier@gmail.com> Signed-off-by:
-
H Hartley Sweeten authored
As mentioned in Documentation/CodingStyle, move EXPORT* macro's to the line immediately after the closing function brace line. Also, move the __initcall() similarly. Signed-off-by:
H Hartley Sweeten <hsweeten@visionengravers.com> Cc: Zach Brown <zach.brown@oracle.com> Cc: Benjamin LaHaise <bcrl@kvack.org> Signed-off-by:
-
Jan Beulich authored
gcc permitting variable length arrays makes the current construct used for BUILD_BUG_ON() useless, as that doesn't produce any diagnostic if the controlling expression isn't really constant. Instead, this patch makes it so that a bit field gets used here. Consequently, those uses where the condition isn't really constant now also need fixing. Note that in the gfp.h, kmemcheck.h, and virtio_config.h cases MAYBE_BUILD_BUG_ON() really just serves documentation purposes - even if the expression is compile time constant (__builtin_constant_p() yields true), the array is still deemed of variable length by gcc, and hence the whole expression doesn't have the intended effect. [akpm@linux-foundation.org: make arch/sparc/include/asm/vio.h compile] [akpm@linux-foundation.org: more nonsensical assertions in tpm.c..] Signed-off-by:
Jan Beulich <jbeulich@novell.com> Cc: Andi Kleen <andi@firstfloor.org> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: "David S. Miller" <davem@davemloft.net> Cc: Rajiv Andrade <srajiv@linux.vnet.ibm.com> Cc: Mimi Zohar <zohar@us.ibm.com> Cc: James Morris <jmorris@namei.org> Signed-off-by:
-
H Hartley Sweeten authored
According to Documentation/CodingStyle the EXPORT* macro should follow immediately after the closing function brace line. Also, mark_buffer_async_write_endio() and do_thaw_all() are not used elsewhere so they should be marked as static. In addition, file_fsync() is actually in fs/sync.c so move the EXPORT* to that file. Signed-off-by:
-
Nick Piggin authored
We have had a report of bad memory allocation latency during DVD-RAM (UDF) writing. This is causing the user's desktop session to become unusable. Jan tracked the cause of this down to UDF inode reclaim blocking: gnome-screens D ffff810006d1d598 0 20686 1 ffff810006d1d508 0000000000000082 ffff810037db6718 0000000000000800 ffff810006d1d488 ffffffff807e4280 ffffffff807e4280 ffff810006d1a580 ffff8100bccbc140 ffff810006d1a8c0 0000000006d1d4e8 ffff810006d1a8c0 Call Trace: [<ffffffff804477f3>] io_schedule+0x63/0xa5 [<ffffffff802c2587>] sync_buffer+0x3b/0x3f [<ffffffff80447d2a>] __wait_on_bit+0x47/0x79 [<ffffffff80447dc6>] out_of_line_wait_on_bit+0x6a/0x77 [<ffffffff802c24f6>] __wait_on_buffer+0x1f/0x21 [<ffffffff802c442a>] __bread+0x70/0x86 [<ffffffff88de9ec7>] :udf:udf_tread+0x38/0x3a [<ffffffff88de0fcf>] :udf:udf_update_inode+0x4d/0x68c [<ffffffff88de26e1>] :udf:udf_write_inode+0x1d/0x2b [<ffffffff802bcf85>] __writeback_single_inode+0x1c0/0x394 [<ffffffff802bd205>] write_inode_now+0x7d/0xc4 [<ffffffff88de2e76>] :udf:udf_clear_inode+0x3d/0x53 [<ffffffff802b39ae>] clear_inode+0xc2/0x11b [<ffffffff802b3ab1>] dispose_list+0x5b/0x102 [<ffffffff802b3d35>] shrink_icache_memory+0x1dd/0x213 [<ffffffff8027ede3>] shrink_slab+0xe3/0x158 [<ffffffff8027fbab>] try_to_free_pages+0x177/0x232 [<ffffffff8027a578>] __alloc_pages+0x1fa/0x392 [<ffffffff802951fa>] alloc_page_vma+0x176/0x189 [<ffffffff802822d8>] __do_fault+0x10c/0x417 [<ffffffff80284232>] handle_mm_fault+0x466/0x940 [<ffffffff8044b922>] do_page_fault+0x676/0xabf This blocks with iprune_mutex held, which then blocks other reclaimers: X D ffff81009d47c400 0 17285 14831 ffff8100844f3728 0000000000000086 0000000000000000 ffff81000000e288 ffff81000000da00 ffffffff807e4280 ffffffff807e4280 ffff81009d47c400 ffffffff805ff890 ffff81009d47c740 00000000844f3808 ffff81009d47c740 Call Trace: [<ffffffff80447f8c>] __mutex_lock_slowpath+0x72/0xa9 [<ffffffff80447e1a>] mutex_lock+0x1e/0x22 [<ffffffff802b3ba1>] shrink_icache_memory+0x49/0x213 [<ffffffff8027ede3>] shrink_slab+0xe3/0x158 [<ffffffff8027fbab>] try_to_free_pages+0x177/0x232 [<ffffffff8027a578>] __alloc_pages+0x1fa/0x392 [<ffffffff8029507f>] alloc_pages_current+0xd1/0xd6 [<ffffffff80279ac0>] __get_free_pages+0xe/0x4d [<ffffffff802ae1b7>] __pollwait+0x5e/0xdf [<ffffffff8860f2b4>] :nvidia:nv_kern_poll+0x2e/0x73 [<ffffffff802ad949>] do_select+0x308/0x506 [<ffffffff802adced>] core_sys_select+0x1a6/0x254 [<ffffffff802ae0b7>] sys_select+0xb5/0x157 Now I think the main problem is having the filesystem block (and do IO) in inode reclaim. The problem is that this doesn't get accounted well and penalizes a random allocator with a big latency spike caused by work generated from elsewhere. I think the best idea would be to avoid this. By design if possible, or by deferring the hard work to an asynchronous context. If the latter, then the fs would probably want to throttle creation of new work with queue size of the deferred work, but let's not get into those details. Anyway, the other obvious thing we looked at is the iprune_mutex which is causing the cascading blocking. We could turn this into an rwsem to improve concurrency. It is unreasonable to totally ban all potentially slow or blocking operations in inode reclaim, so I think this is a cheap way to get a small improvement. This doesn't solve the whole problem of course. The process doing inode reclaim will still take the latency hit, and concurrent processes may end up contending on filesystem locks. So fs developers should keep these problems in mind. Signed-off-by:
Nick Piggin <npiggin@suse.de> Cc: Jan Kara <jack@ucw.cz> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Christoph Hellwig <hch@lst.de> Signed-off-by:
-
