debugfs: dont stop on first failed recursive delete
While running a while loop of removing a module that removes a debugfs
directory with debugfs_remove_recursive, and at the same time doing a
while loop of cat of a file in that directory, I would hit a point where
somehow the cat of the file caused the remove to fail.
The result is that other files did not get removed when the module
was removed. I simple read of one of those file can oops the kernel
because the operations to the file no longer exist (removed by module).
The funny thing is that the file being cat'ed was removed. It was
the siblings that were not. I see in the code to debugfs_remove_recursive
there's a test that checks if the child fails to bail out of the loop
to prevent an infinite loop.
What this patch does is to still try any siblings in that directory.
If all the siblings fail, or there are no more siblings, then we exit
the loop.
This fixes the above symptom, but...
This is no full proof. It makes the debugfs_remove_recursive a bit more
robust, but it does not explain why the one file failed. There may
be some kind of delay deletion that makes the debugfs think it did
not succeed. So this patch is more of a fix for the symptom but not
the disease.
This patch still makes the debugfs_remove_recursive more robust and
until I can find out why the bug exists, this patch will keep
the kernel from oopsing in most cases. Even after the cause is found
I think this change can stand on its own and should be kept.
[ Impact: prevent kernel oops on module unload and reading debugfs files ]
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This converts resource and IRQ getbyname functions for the platform
bus to use const char *, I ran into compiler moanings when I tried
using a const char * for looking up a certain resource.
Signed-off-by: Linus Walleij <linus.walleij@stericsson.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch adds a new bus notifier event which is emitted _after_ a
device is removed from its driver. This event will be used by the
dma-api debug code to check if a driver has released all dma allocations
for that device.
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
There is the possiblity of a memory leak if a page is allocated and if
sysfs_getlink() fails in the sysfs_follow_link.
Signed-off-by: Armin Kuster <akuster@mvista.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Add a blurb to the driver-model documentation about how (not) to add
extra attributes to a struct device at driver probe time.
Signed-off-by: Grant Likely <grant.likely@secretlab.ca>
Cc: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
During bootup performance tracing we see repeated occurrences of
/sys/kernel/uid/* events for the same uid, leading to a,
in this case, rather pointless userspace processing for the
same uid over and over.
This is usually caused by tools which change their uid to "nobody",
to run without privileges to read data supplied by untrusted users.
This change delays the execution of the (already existing) scheduled
work, to cleanup the uid after one second, so the allocated and announced
uid can possibly be re-used by another process.
This is the current behavior, where almost every invocation of a
binary, which changes the uid, creates two events:
$ read START < /sys/kernel/uevent_seqnum; \
for i in `seq 100`; do su --shell=/bin/true bin; done; \
read END < /sys/kernel/uevent_seqnum; \
echo $(($END - $START))
178
With the delayed cleanup, we get only two events, and userspace finishes
a bit faster too:
$ read START < /sys/kernel/uevent_seqnum; \
for i in `seq 100`; do su --shell=/bin/true bin; done; \
read END < /sys/kernel/uevent_seqnum; \
echo $(($END - $START))
1
Acked-by: Dhaval Giani <dhaval@linux.vnet.ibm.com>
Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
All recent distros depend on the non-deprecated sysfs layout, so
change the default value of the option to reflect that.
Signed-off-by: Kay Sievers <kay.sievers@vrfy.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit_fs_roots skips updating root items for fs trees that aren't modified.
This is unsafe now that relocation code modifies root item's last_snapshot
field without modifying corresponding fs tree.
Signed-off-by: Yan Zheng <zheng.yan@oracle.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
Replace wait_event_interruptible() with wait_event() in o2net_send_message_vec().
This is because this function is called by the dlm that expects signals to be
blocked.
Fixes oss bugzilla#1126
http://oss.oracle.com/bugzilla/show_bug.cgi?id=1126
Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
In ocfs2_add_branch, we use the rightmost rec of the leaf extent block
to generate the e_cpos for the newly added branch. In the most case, it
is OK but if the parent extent block's rightmost rec covers more clusters
than the leaf does, it will cause kernel panic if we insert some clusters
in it. The message is something like:
(7445,1):ocfs2_insert_at_leaf:3775 ERROR: bug expression:
le16_to_cpu(el->l_next_free_rec) >= le16_to_cpu(el->l_count)
(7445,1):ocfs2_insert_at_leaf:3775 ERROR: inode 66053, depth 0, count 28,
next free 28, rec.cpos 270, rec.clusters 1, insert.cpos 275, insert.clusters 1
[<fa7ad565>] ? ocfs2_do_insert_extent+0xb58/0xda0 [ocfs2]
[<fa7b08f2>] ? ocfs2_insert_extent+0x5bd/0x6ba [ocfs2]
[<fa7b1b8b>] ? ocfs2_add_clusters_in_btree+0x37f/0x564 [ocfs2]
...
The panic can be easily reproduced by the following small test case
(with bs=512, cs=4K, and I remove all the error handling so that it looks
clear enough for reading).
int main(int argc, char **argv)
{
int fd, i;
char buf[5] = "test";
fd = open(argv[1], O_RDWR|O_CREAT);
for (i = 0; i < 30; i++) {
lseek(fd, 40960 * i, SEEK_SET);
write(fd, buf, 5);
}
ftruncate(fd, 1146880);
lseek(fd, 1126400, SEEK_SET);
write(fd, buf, 5);
close(fd);
return 0;
}
The reason of the panic is that:
the 30 writes and the ftruncate makes the file's extent list looks like:
Tree Depth: 1 Count: 19 Next Free Rec: 1
## Offset Clusters Block#
0 0 280 86183
SubAlloc Bit: 7 SubAlloc Slot: 0
Blknum: 86183 Next Leaf: 0
CRC32: 00000000 ECC: 0000
Tree Depth: 0 Count: 28 Next Free Rec: 28
## Offset Clusters Block# Flags
0 0 1 143368 0x0
1 10 1 143376 0x0
...
26 260 1 143576 0x0
27 270 1 143584 0x0
Now another write at 1126400(275 cluster) whiich will write at the gap
between 271 and 280 will trigger ocfs2_add_branch, but the result after
the function looks like:
Tree Depth: 1 Count: 19 Next Free Rec: 2
## Offset Clusters Block#
0 0 280 86183
1 271 0 143592
So the extent record is intersected and make the following operation bug out.
This patch just try to remove the gap before we add the new branch, so that
the root(branch) rightmost rec will cover the same right position. So in the
above case, before adding branch the tree will be changed to
Tree Depth: 1 Count: 19 Next Free Rec: 1
## Offset Clusters Block#
0 0 271 86183
SubAlloc Bit: 7 SubAlloc Slot: 0
Blknum: 86183 Next Leaf: 0
CRC32: 00000000 ECC: 0000
Tree Depth: 0 Count: 28 Next Free Rec: 28
## Offset Clusters Block# Flags
0 0 1 143368 0x0
1 10 1 143376 0x0
...
26 260 1 143576 0x0
27 270 1 143584 0x0
And after branch add, the tree looks like
Tree Depth: 1 Count: 19 Next Free Rec: 2
## Offset Clusters Block#
0 0 271 86183
1 271 0 143592
Signed-off-by: Tao Ma <tao.ma@oracle.com>
Acked-by: Mark Fasheh <mfasheh@suse.com>
Signed-off-by: Joel Becker <joel.becker@oracle.com>
* 'timers-for-linus-migration' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
timers: Logic to move non pinned timers
timers: /proc/sys sysctl hook to enable timer migration
timers: Identifying the existing pinned timers
timers: Framework for identifying pinned timers
timers: allow deferrable timers for intervals tv2-tv5 to be deferred
Fix up conflicts in kernel/sched.c and kernel/timer.c manually
* 'timers-for-linus-clockevents' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
clockevent: export register_device and delta2ns
clockevents: tick_broadcast_device can become static
* 'timers-for-linus-clocksource' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
clocksource: prevent selection of low resolution clocksourse also for nohz=on
clocksource: sanity check sysfs clocksource changes
Export the alarm flags provided by the MAX6650/MAX6651 fan-speed regulator
and monitor chips via sysfs.
Signed-off-by: Christian Engelmayer <christian.engelmayer@frequentis.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Add support for the hwmon part of the Fintek F71858FG superio IC to the
f71882fg driver. Many thanks to Jelle de Jong for lending me a motherboard
with this superio on it.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
While working on f71852fg support I noticed that the f8000 sysfs attr
table was missing entries for temp#_fault, which the f8000 does have and
which we were already reading.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
We depend up on the pwm_enable register (0x96) not containing any reserved
settings in various places. We were already checking to make sure there
were no reserved settings in the register for the f71862fg, this patch adds
the same checking for the f8000, while at it it also moves the code to
a more apropriate place so we don't need to check if the fan/pwm part
of the IC is enabled twice.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Currently we are using separate per model sysfs attr for the 3th pwm, because
the 3th pwm of the f8000 only has automatic mode and not manual mode. Doing
things this way was getting in the way for adding f71858fg support, so this
patch makes the pwm attr identical for all models, and instead adds a check
to store_pwm_enable() disallowing setting the 3th pwm to manual mode
on a f8000 IC.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
The MSI MS-7031 is based on an ATI IXP300 south bridge. On this south
bridge, accessible I/O ports must be enabled explicitly. Unfortunately
the BIOS forgets to enable access to the hardware monitoring chip I/O
ports, so hardware monitoring fails.
Add a quirk enabling access to the required ports (0x295-0x296). This
is exactly what MSI's own hardware monitoring application is doing, so
it has to be the right way.
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Add support for the new incarnation of the Winbond/Nuvoton W83627DHG
chip known as W83627DHG-P. It is basically the same as the original
W83627DHG with an additional automatic can speed control mode (not
supported by the driver yet.)
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Tested-by: Madhu <madhu.chinakonda@gmail.com>
Documentation for the tmp401 driver.
The documentation describes the tmp401 driver and the supported Texas
Instruments TMP401 and TMP411 temperature sensor chips.
Further documentation for new sysfs attributes supported by this
driver is added to Documentation/hwmon/sysfs-interface.
Signed-off-by: Andre Prendel <andre.prendel@gmx.de>
Acked-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
This adds support for TI's TMP411 sensor chip.
Preliminary support were done by Gabriel Konat, Sander Leget and
Wouter Willems. The chip is compatible with TI's TMP401 sensor
chip. It has additional support for historical minimun/maximum
measurements.
Signed-off-by: Andre Prendel <andre.prendel@gmx.de>
Acked-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
This is a new hwmon driver for TI's TMP401 temperature sensor IC. This driver
was written on behalf of an embedded systems vendor under the
Linux driver project.
It has been tested using a TI TMP401 sample attached to a i2c-tiny-usb adapter.
Which was provided by Till Harbaum, many thanks to him for this!
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Enable auto-probing for the HC10 blade and amend the supported system
list.
Signed-off-by: Darrick J. Wong <djwong@us.ibm.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Make sure __devexit and devexit_p() match in all hwmon drivers.
Suggested by a similar fix from Mike Frysinger.
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Acked-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Jonathan Cameron <jic23@cam.ac.uk>
* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc: (103 commits)
powerpc: Fix bug in move of altivec code to vector.S
powerpc: Add support for swiotlb on 32-bit
powerpc/spufs: Remove unused error path
powerpc: Fix warning when printing a resource_size_t
powerpc/xmon: Remove unused variable in xmon.c
powerpc/pseries: Fix warnings when printing resource_size_t
powerpc: Shield code specific to 64-bit server processors
powerpc: Separate PACA fields for server CPUs
powerpc: Split exception handling out of head_64.S
powerpc: Introduce CONFIG_PPC_BOOK3S
powerpc: Move VMX and VSX asm code to vector.S
powerpc: Set init_bootmem_done on NUMA platforms as well
powerpc/mm: Fix a AB->BA deadlock scenario with nohash MMU context lock
powerpc/mm: Fix some SMP issues with MMU context handling
powerpc: Add PTRACE_SINGLEBLOCK support
fbdev: Add PLB support and cleanup DCR in xilinxfb driver.
powerpc/virtex: Add ml510 reference design device tree
powerpc/virtex: Add Xilinx ML510 reference design support
powerpc/virtex: refactor intc driver and add support for i8259 cascading
powerpc/virtex: Add support for Xilinx PCI host bridge
...
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/lrg/voltage-2.6:
regulator/max1586: fix V3 gain calculation integer overflow
regulator/max1586: support increased V3 voltage range
regulator: lp3971 - fix driver link error when built-in.
LP3971 PMIC regulator driver (updated and combined version)
regulator: remove driver_data direct access of struct device
regulator: Set MODULE_ALIAS for regulator drivers
regulator: Support list_voltage for fixed voltage regulator
regulator: Move regulator drivers to subsys_initcall()
regulator: build fix for powerpc - renamed show_state
regulator: add userspace-consumer driver
Maxim 1586 regulator driver
ad6561dffa ("module: trim exception table on init
free.") put a bogus trim_init_extable() function into ia64 which didn't compile.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Tony Luck <tony.luck@intel.com>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ryusuke/nilfs2: (22 commits)
nilfs2: support contiguous lookup of blocks
nilfs2: add sync_page method to page caches of meta data
nilfs2: use device's backing_dev_info for btree node caches
nilfs2: return EBUSY against delete request on snapshot
nilfs2: modify list of unsupported features in caveats
nilfs2: enable sync_page method
nilfs2: set bio unplug flag for the last bio in segment
nilfs2: allow future expansion of metadata read out via get info ioctl
NILFS2: Pagecache usage optimization on NILFS2
nilfs2: remove nilfs_btree_operations from btree mapping
nilfs2: remove nilfs_direct_operations from direct mapping
nilfs2: remove bmap pointer operations
nilfs2: remove useless b_low and b_high fields from nilfs_bmap struct
nilfs2: remove pointless NULL check of bpop_commit_alloc_ptr function
nilfs2: move get block functions in bmap.c into btree codes
nilfs2: remove nilfs_bmap_delete_block
nilfs2: remove nilfs_bmap_put_block
nilfs2: remove header file for segment list operations
nilfs2: eliminate removal list of segments
nilfs2: add sufile function that can modify multiple segment usages
...
The legacy i2c binding model is going away soon, so convert the ppc
therm_windtunnel driver to the new model or it will break.
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
The legacy i2c binding model is going away soon, so convert the ppc
therm_adt746x driver to the new model or it will break.
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Tested-by: Andreas Schwab <schwab@linux-m68k.org>
Tested-by: Paul Mackerras <paulus@samba.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The legacy i2c binding model is going away soon, so convert the
macintosh windfarm drivers to the new model or they will break.
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Tested-by: Johannes Berg <johannes@sipsolutions.net>
Tested-by: Paul Mackerras <paulus@samba.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The legacy i2c binding model is going away soon, so convert the
macintosh therm_pm72 driver to the new model or it will break.
This is really a quick and dirty conversion, that should do the trick
for now, but no doubt that something cleaner can be done if anyone is
interested.
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Tested-by: Paul Mackerras <paulus@samba.org>
The south bridge of the VIA VX855 chipset has a different PCI Device ID
so i2c-viapro.c needs to be updated with this.
Signed-off-by: Harald Welte <HaraldWelte@viatech.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
This driver only reads the user EEPROM of that chip, so we can move it
to the eeprom-directory in order to further clean up (and later remove)
drivers/i2c/chips.
The Kconfig text was updated to match the current functionality,
dropping the meanwhile obsoleted parts.
Defconfigs have been adapted.
Signed-off-by: Wolfram Sang <w.sang@pengutronix.de>
Acked-by: Ben Gardner <gardner.ben@gmail.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
We don't need to give adapters a parent if they don't have one. The
driver core will put them in the virtual device directory and all will
be fine.
Signed-off-by: Jean Delvare <khali@linux-fr.org>
There's no point in giving the I2C bus of Voodoo3 adapters a class
value, there's no video chip driver checking for it anymore. If
support is ever needed, the video device should be instantiated
explicitly rather than probed.
To the best of my knowledge the only video chip that can be found on
these boards is a BT869 video encoder, for which no support exists
currently.
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Acked-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Some small changes in i2c core to retry i2c xfers until either the
maximum number of retries or the timeout is hit.
Signed-off-by: Clifford Wolf <clifford@clifford.at>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
I just realized this has a kmap_atomic bug in...
The below would fix it - but it's complicating this code
some more.
Alternatively I would have to introduce something like
pte_offset_map_irq() which would make the irq/nmi detection and leave
the regular code paths alone, however that would mean either duplicating
the gup_fast() pagewalk or passing down a pte function pointer, which
would only duplicate the gup_pte_range() bit, neither is really
attractive ...
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
CC: Nick Piggin <npiggin@suse.de>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Yong Wang reported the following compiler warning:
builtin-report.c: In function 'process_overflow_event':
builtin-report.c:984: error: cast to pointer from integer of different size
Which happens because we try to print ->ips[] out with a limited
format, losing the high 32 bits. Print it out using %016Lx instead.
Reported-by: Yong Wang <yong.y.wang@linux.intel.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
At present, every architecture that supports perf_counters has to
declare set_perf_counter_pending() in its arch-specific headers.
This consolidates the declarations into a single declaration in one
common place, include/linux/perf_counter.h. On powerpc, we continue
to provide a static inline definition of set_perf_counter_pending()
in the powerpc hw_irq.h.
Also, this removes from the x86 perf_counter.h the unused null
definitions of {test,clear}_perf_counter_pending.
Reported-by: Mike Frysinger <vapier.adi@gmail.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: benh@kernel.crashing.org
LKML-Reference: <18998.13388.920691.523227@cargo.ozlabs.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This fixes a couple of compile warnings that crept into the powerpc
perf_counter code recently:
CC arch/powerpc/kernel/perf_counter.o
arch/powerpc/kernel/perf_counter.c: In function 'record_and_restart':
arch/powerpc/kernel/perf_counter.c:1016: warning: unused variable 'addr'
arch/powerpc/kernel/perf_counter.c: In function 'hw_perf_counter_init':
arch/powerpc/kernel/perf_counter.c:891: warning: 'ev' may be used uninitialized in this function
Stephen Rothwell reported this against linux-next as well.
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <18998.12884.787039.22202@cargo.ozlabs.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Take advantage of call-graph percounter sampling/recording to
display a non-trivial histogram: the true, collapsed/summarized
cost measurement, on a per system call total overhead basis:
aldebaran:~/linux/linux/tools/perf> ./perf record -g -a -f ~/hackbench 10
aldebaran:~/linux/linux/tools/perf> ./perf report -s symbol --syscalls | head -10
#
# (3536 samples)
#
# Overhead Symbol
# ........ ......
#
40.75% [k] sys_write
40.21% [k] sys_read
4.44% [k] do_nmi
...
This is done by accounting each (reliable) call-chain that chains back
to a given system call to that system call function.
[ So in the above example we can see that hackbench spends about 40% of
its total time somewhere in sys_write() and 40% somewhere in
sys_read(), the rest of the time is spent in user-space. The time
is not spent in sys_write() _itself_ but in one of its many child
functions. ]
Or, a recording of a (source files are already in the page-cache) kernel build:
$ perf record -g -m 512 -f -- make -j32 kernel
$ perf report -s s --syscalls | grep '\[k\]' | grep -v nmi
4.14% [k] do_page_fault
1.20% [k] sys_write
1.10% [k] sys_open
0.63% [k] sys_exit_group
0.48% [k] smp_apic_timer_interrupt
0.37% [k] sys_read
0.37% [k] sys_execve
0.20% [k] sys_mmap
0.18% [k] sys_close
0.14% [k] sys_munmap
0.13% [k] sys_poll
0.09% [k] sys_newstat
0.07% [k] sys_clone
0.06% [k] sys_newfstat
0.05% [k] sys_access
0.05% [k] schedule
Shows the true total cost of each syscall variant that gets used
during a kernel build. This profile reveals it that pagefaults are
the costliest, followed by read()/write().
An interesting detail: timer interrupts cost 0.5% - or 0.5 seconds
per 100 seconds of kernel build-time. (this was done with HZ=1000)
The summary is done in 'perf report', i.e. in the post-processing
stage - so once we have a good call-graph recording, this type of
non-trivial high-level analysis becomes possible.
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
__copy_from_user_inatomic() isn't NMI safe in that it can trigger
the page fault handler which is another trap and its return path
invokes IRET which will also close the NMI context.
Therefore use a GUP based approach to copy the stack frames over.
We tried an alternative solution as well: we used a forward ported
version of Mathieu Desnoyers's "NMI safe INT3 and Page Fault" patch
that modifies the exception return path to use an open-coded IRET with
explicit stack unrolling and TF checking.
This didnt work as it interacted with faulting user-space instructions,
causing them not to restart properly, which corrupts user-space
registers.
Solving that would probably involve disassembling those instructions
and backtracing the RIP. But even without that, the code was deemed
rather complex to the already non-trivial x86 entry assembly code,
so instead we went for this GUP based method that does a
software-walk of the pagetables.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Cc: Vegard Nossum <vegard.nossum@gmail.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
