Mugunthan V N [Mon, 29 Apr 2013 23:18:17 +0000 (16:18 -0700)]
checkpatch: add Suggested-by as a standard signature
As people started using Suggested-by as standard signature, adding
"Suggested-by" to the standard signature so that checkpatch won't
generate warning when Suggested-by is used in patch signature
checkpatch: only warn for empty lines before closing braces by themselves
This check was intended to catch extra newlines at the end of a function
definition, but it would trigger on any closing brace, including those
of inline functions and macro definitions, triggering false positives.
Now, only closing braces on a line by themselves trigger this check.
Tested with:
$ cat test.h
/* test.h - Test file */
static inline int foo(void) { return 0; }
static inline int bar(void)
{
return 1;
}
$ ./scripts/checkpatch.pl --strict -f test.h # Before this commit
CHECK: Blank lines aren't necessary before a close brace '}'
+
+static inline int foo(void) { return 0; }
CHECK: Blank lines aren't necessary before a close brace '}'
+
+}
Joe Perches [Mon, 29 Apr 2013 23:18:12 +0000 (16:18 -0700)]
checkpatch: add check for reuse of krealloc arg
On Thu, 2013-03-14 at 13:30 +0000, David Woodhouse wrote:
> If krealloc() returns NULL, it *doesn't* free the original. So any code
> of the form 'foo = krealloc(foo, …);' is almost certainly a bug.
argv_split() allocates argv[count_argc(str)] array and assumes that it
will find the same number of arguments later. This is obviously wrong if
this string can be changed, say, by sysctl.
With this patch argv_split() kstrndup's the whole string and does not
split it, we simply replace the spaces with zeroes and keep the allocated
memory in argv[-1] for argv_free(arg).
We do not use argv[0] because:
- str can be all-spaces or empty. In fact this case is fine,
we could kfree() it before return, but:
- str can have a space at the start, and we can not rely on
kstrndup(skip_spaces(str)) because it can equally race if
this string is mutable.
Also, simplify count_argc() and kill the no longer used skip_arg().
Kim, Milo [Mon, 29 Apr 2013 23:18:06 +0000 (16:18 -0700)]
backlight: lp855x: add a device tree structure
Enable supporting the DT structure of LP855x family devices. If the
platform data is NULL, the driver tries to parse a DT structure. Then,
the platform data is copied from the DT. Documentation is added as well.
Kim, Milo [Mon, 29 Apr 2013 23:18:05 +0000 (16:18 -0700)]
backlight: lp855x: remove duplicate platform data
The 'load_new_rom_data' was used for checking whether new ROM data should
be updated or not.
However, we can decide it with 'size_program' data. If the size is
greater than 0, it means updating ROM area is required. Otherwise, the
default ROM data will be used. Therefore, this duplicate platform data
can be removed.
Kim, Milo [Mon, 29 Apr 2013 23:18:03 +0000 (16:18 -0700)]
backlight: lp855x: move backlight mode platform data
The brightness of LP855x devices is controlled by I2C register or PWM
input . This mode was selected through the platform data, but it can be
chosen by the driver internally without platform data configuration.
How to decide the control mode:
If the PWM period has specific value, the mode is PWM input.
On the other hand, the mode is register-based.
This mode selection is done on the _probe().
Move 'mode' from a header file to the driver private data structure,
'lp855 x'. And correlated code was replaced.
Jingoo Han [Mon, 29 Apr 2013 23:18:01 +0000 (16:18 -0700)]
backlight: vgg2432a4: convert vgg2432a4_driver to dev_pm_ops
Instead of using legacy suspend/resume methods, using newer dev_pm_ops
structure allows better control over power management. Also, use of
pm_message_t is deprecated. Thus, it is removed.
Platform LCD devices may need to do some device-specific initialization
before they can be used (regulator or GPIO setup, for example), but
currently the driver does not support any way of doing this. This patch
adds a probe() callback to plat_lcd_data which platform LCD devices can
set to indicate that device-specific initialization is needed.
Jingoo Han [Mon, 29 Apr 2013 23:17:57 +0000 (16:17 -0700)]
backlight: ili922x: use spi_set_drvdata()
Use the wrapper functions for getting and setting the driver data using
spi_device instead of using dev_{get|set}_drvdata with &spi->dev, so we
can directly pass a struct spi_device.
Jingoo Han [Mon, 29 Apr 2013 23:17:56 +0000 (16:17 -0700)]
backlight: ili9320: use spi_set_drvdata()
Use the wrapper functions for getting and setting the driver data using
spi_device instead of using dev_{get|set}_drvdata with &spi->dev, so we
can directly pass a struct spi_device.
Add LCD driver for Ilitek ILI9221/ILI9222 controller. The driver uses
SPI interface for controller access and configuration and RGB interface
for graphics data transfer.
drivers/video/backlight/adp5520_bl.c: fix compiler warning in adp5520_show()
While compiling with make W=1 (gcc gcc (GCC) 4.7.2 20121109 (Red Hat
4.7.2-8)) the following warning happens:
drivers/video/backlight/adp5520_bl.c: In function `adp5520_show':
drivers/video/backlight/adp5520_bl.c:146:6: warning: variable `error' set but not used [-Wunused-but-set-variable]
fixed by checking the return value of the variable
Paul Bolle [Mon, 29 Apr 2013 23:17:48 +0000 (16:17 -0700)]
drivers/video/backlight/Kconfig: fix typo "MACH_SAM9...EK" three times
Fix three typos (originally) introduced by commit a9a84c37d1ee
("atmel_lcdfb: backlight control").
Two of these typos were introduced in v2.6.25. (The third was
introduced in 915190f7d4f0 ("[ARM] 5614/1: at91: atmel_lcdfb: add
at91sam9g10 support to atmel LCD driver")). Checking these commits
reveals that the default value of 'y' has never been set automatically
in all releases since v2.6.25! Perhaps this line might as well be
dropped.
Jingoo Han [Mon, 29 Apr 2013 23:17:34 +0000 (16:17 -0700)]
drivers/video/backlight/l4f00242t03.c: check return value of regulator_enable()
regulator_enable() is marked as as __must_check. Therefore the return
value of regulator_enable() should be checked. Also, this patch checks
return value of regulator_set_voltage().
Add missing braces to include error message. The error message is
related to the return value for sysfs_create_group(). However,
sysfs_create_group() is called when pdata->en_ambl_sens is not zero.
Thus, the checking return value should be included in the if statement.
drivers/video/backlight/lp855x_bl.c: fix compiler warning in lp855x_probe
While doing with make W=1 gcc (gcc (GCC) 4.7.2 20121109 (Red Hat
4.7.2-8)) I found
drivers/video/backlight/lp855x_bl.c: In function `lp855x_probe':
drivers/video/backlight/lp855x_bl.c:342:35: warning: variable `mode' set but not used [-Wunused-but-set-variable]
fixed by removing it as since its not used anywhere
When the macro such as SIMPLE_DEV_PM_OPS is used, there is no need to
use '#ifdef CONFIG_PM' to prevent build error. Thus, this patch removes
unnecessary ifdefs.
Stephen Warren [Mon, 29 Apr 2013 23:17:23 +0000 (16:17 -0700)]
get_maintainer: use filename-only regex match for Tegra
Create a new N: entry type in MAINTAINERS which performs a regex match
against filenames; either those extracted from patch +++ or --- lines,
or those specified on the command-line using the -f option.
This provides the same benefits as using a K: regex option to match a
set of filenames (see commit eb90d0855b75 "get_maintainer: allow
keywords to match filenames"), but without the disadvantage that
"random" file content, such as comments, will ever match the regex.
Hence, revert most of that commit.
Andrew Morton [Mon, 29 Apr 2013 23:17:19 +0000 (16:17 -0700)]
include/linux/printk.h: include stdarg.h
printk.h uses va_list but doesn't include stdarg.h. Hence printk.h is
unusable unless its includer has already included kernel.h (which includes
stdarg.h).
Remove the dependency by including stdarg.h in printk.h
Commit 7ff9554bb578 ("printk: convert byte-buffer to variable-length
record buffer") removed start and end parameters from
call_console_drivers, but those parameters still exist in
include/trace/events/printk.h.
Without start and end parameters handling, printk tracing became more
simple as: trace_console(text, len);
gcc4: disable __compiletime_object_size for GCC 4.6+
__builtin_object_size is known to be broken on gcc 4.6+.
See http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48880 for details.
This causes unnecssary build warnings and errors such as
In function 'copy_from_user', inlined from 'sb16_copy_from_user'
at sound/oss/sb_audio.c:878:22:
arch/x86/include/asm/uaccess_32.h:211:26: error: call to 'copy_from_user_overflow'
declared with attribute error: copy_from_user() buffer size is not provably correct
make[3]: [sound/oss/sb_audio.o] Error 1 (ignored)
Philipp Zabel [Mon, 29 Apr 2013 23:17:14 +0000 (16:17 -0700)]
media: coda: use genalloc API
This patch depends on "genalloc: add devres support, allow to find a
managed pool by device", which provides the of_get_named_gen_pool and
dev_get_gen_pool functions.
Philipp Zabel [Mon, 29 Apr 2013 23:17:12 +0000 (16:17 -0700)]
misc: generic on-chip SRAM allocation driver
This driver requests and remaps a memory region as configured in the
device tree. It serves memory from this region via the genalloc API. It
optionally enables the SRAM clock.
Other drivers can retrieve the genalloc pool from a phandle pointing to
this drivers' device node in the device tree.
The allocation granularity is hard-coded to 32 bytes for now, to make the
SRAM driver useful for the 6502 remoteproc driver. There is overhead for
bigger SRAMs, where only a much coarser allocation granularity is needed:
At 32 bytes minimum allocation size, a 256 KiB SRAM needs a 1 KiB bitmap
to track allocations.
Philipp Zabel [Mon, 29 Apr 2013 23:17:10 +0000 (16:17 -0700)]
genalloc: add devres support, allow to find a managed pool by device
This patch adds three exported functions to lib/genalloc.c:
devm_gen_pool_create, dev_get_gen_pool, and of_get_named_gen_pool.
devm_gen_pool_create is a managed version of gen_pool_create that keeps
track of the pool via devres and allows the management code to
automatically destroy it after device removal.
dev_get_gen_pool retrieves the gen_pool for a given device, if it was
created with devm_gen_pool_create, using devres_find.
of_get_named_gen_pool retrieves the gen_pool for a given device node and
property name, where the property must contain a phandle pointing to a
platform device node. The corresponding platform device is then fed into
dev_get_gen_pool and the resulting gen_pool is returned.
The reason why the messages are shown is to release a resource
structure, allocated by bootmem, by kfree(). So when we release a
resource structure, we should check whether it is allocated by bootmem
or not.
But even if we know a resource structure is allocated by bootmem, we
cannot release it since SLxB cannot treat it. So for reusing a resource
structure, this patch remembers it by using bootmem_resource as follows:
When releasing a resource structure by free_resource(), free_resource()
checks whether the resource structure is allocated by bootmem or not.
If it is allocated by bootmem, free_resource() adds it to
bootmem_resource. If it is not allocated by bootmem, free_resource()
release it by kfree().
And when getting a new resource structure by get_resource(),
get_resource() checks whether bootmem_resource has released resource
structures or not. If there is a released resource structure,
get_resource() returns it. If there is not a releaed resource
structure, get_resource() returns new resource structure allocated by
kzalloc().
There are times when HIGHMEM is enabled, but we don't prefer
CONFIG_BOUNCE to be enabled. CONFIG_BOUNCE can reduce the block device
throughput, and this is not ideal for machines where we don't gain much
by enabling it. So provide an option to deselect CONFIG_BOUNCE. The
observation was made while measuring eMMC throughput using iozone on an
ARM device with 1GB RAM.
numa, cpu hotplug: change links of CPU and node when changing node number by onlining CPU
When booting x86 system contains memoryless node, node numbers of CPUs
on memoryless node were changed to nearest online node number by
init_cpu_to_node() because the node is not online.
In my system, node numbers of cpu#30-44 and 75-89 were changed from 2 to
0 as follows:
If we hot add memory to memoryless node and offine/online all CPUs on
the node, node numbers of these CPUs are changed to correct node numbers
by srat_detect_node() because the node become online.
In this case, node numbers of cpu#30-44 and 75-89 were changed from 0 to
2 in my system as follows:
Randy Dunlap [Mon, 29 Apr 2013 22:08:49 +0000 (15:08 -0700)]
mm: fix memory_hotplug.c printk format warning
PFN_PHYS() is a phys_addr_t, which can be u32 or u64.
Fix the build warning when phys_addr_t is u32.
mm/memory_hotplug.c: warning: format '%llx' expects argument of type 'long long unsigned int', but argument 2 has type 'unsigned int' [-Wformat]: => 1685:3
mm/memory_hotplug.c: warning: format '%llx' expects argument of type 'long long unsigned int', but argument 3 has type 'unsigned int' [-Wformat]: => 1685:3
mm: swap: mark swap pages writeback before queueing for direct IO
As pointed out by Andrew Morton, the swap-over-NFS writeback is not
setting PageWriteback before it is queued for direct IO. While swap
pages do not participate in BDI or process dirty accounting and the IO
is synchronous, the writeback bit is still required and not setting it
in this case was an oversight. swapoff depends on the page writeback to
synchronoise all pending writes on a swap page before it is reused.
Swapcache freeing and reuse depend on checking the PageWriteback under
lock to ensure the page is safe to reuse.
Direct IO handlers and the direct IO handler for NFS do not deal with
PageWriteback as they are synchronous writes. In the case of NFS, it
schedules pages (or a page in the case of swap) for IO and then waits
synchronously for IO to complete in nfs_direct_write(). It is
recognised that this is a slowdown from normal swap handling which is
asynchronous and uses a completion handler. Shoving PageWriteback
handling down into direct IO handlers looks like a bad fit to handle the
swap case although it may have to be dealt with some day if swap is
converted to use direct IO in general and bmap is finally done away
with. At that point it will be necessary to refit asynchronous direct
IO with completion handlers onto the swap subsystem.
As swapcache currently depends on PageWriteback to protect against
races, this patch sets PageWriteback under the page lock before queueing
it for direct IO. It is cleared when the direct IO handler returns. IO
errors are treated similarly to the direct-to-bio case except PageError
is not set as in the case of swap-over-NFS, it is likely to be a
transient error.
It was asked what prevents such a page being reclaimed in parallel.
With this patch applied, such a page will now be skipped (most of the
time) or blocked until the writeback completes. Reclaim checks
PageWriteback under the page lock before calling try_to_free_swap and
the page lock should prevent the page being requeued for IO before it is
freed.
This and Jerome's related patch should considered for -stable as far
back as 3.6 when swap-over-NFS was introduced.
swap: redirty page if page write fails on swap file
Since commit 62c230bc1790 ("mm: add support for a filesystem to activate
swap files and use direct_IO for writing swap pages"), swap_writepage()
calls direct_IO on swap files. However, in that case the page isn't
redirtied if I/O fails, and is therefore handled afterwards as if it has
been successfully written to the swap file, leading to memory corruption
when the page is eventually swapped back in.
This patch sets the page dirty when direct_IO() fails. It fixes a
memory corruption that happened while using swap-over-NFS.
David Rientjes [Mon, 29 Apr 2013 22:08:45 +0000 (15:08 -0700)]
mm, memcg: give exiting processes access to memory reserves
A memcg may livelock when oom if the process that grabs the hierarchy's
oom lock is never the first process with PF_EXITING set in the memcg's
task iteration.
The oom killer, both global and memcg, will defer if it finds an
eligible process that is in the process of exiting and it is not being
ptraced. The idea is to allow it to exit without using memory reserves
before needlessly killing another process.
This normally works fine except in the memcg case with a large number of
threads attached to the oom memcg. In this case, the memcg oom killer
only gets called for the process that grabs the hierarchy's oom lock;
all others end up blocked on the memcg's oom waitqueue. Thus, if the
process that grabs the hierarchy's oom lock is never the first
PF_EXITING process in the memcg's task iteration, the oom killer is
constantly deferred without anything making progress.
The fix is to give PF_EXITING processes access to memory reserves so
that we've marked them as oom killed without any iteration. This allows
__mem_cgroup_try_charge() to succeed so that the process may exit. This
makes the memcg oom killer exemption for TIF_MEMDIE tasks, now
immediately granted for processes with pending SIGKILLs and those in the
exit path, to be equivalent to what is done for the global oom killer.
thp: fix huge zero page logic for page with pfn == 0
Current implementation of huge zero page uses pfn value 0 to indicate
that the page hasn't allocated yet. It assumes that buddy page
allocator can't return page with pfn == 0.
Let's rework the code to store 'struct page *' of huge zero page, not
its pfn. This way we can avoid the weak assumption.
When hot removing memory, a firmware_map_entry which has memory range of
the memory is released by release_firmware_map_entry(). If the entry is
allocated by bootmem, release_firmware_map_entry() adds the entry to
map_entires_bootmem list when firmware_map_find_entry() finds the entry
from map_entries list. But firmware_map_find_entry never find the entry
sicne map_entires list does not have the entry. So the entry just
leaks.
Here are steps of leaking firmware_map_entry:
firmware_map_remove()
-> firmware_map_find_entry()
Find released entry from map_entries list
-> firmware_map_remove_entry()
Delete the entry from map_entries list
-> remove_sysfs_fw_map_entry()
...
-> release_firmware_map_entry()
-> firmware_map_find_entry()
Find the entry from map_entries list but the entry has been
deleted from map_entries list. So the entry is not added
to map_entries_bootmem. Thus the entry leaks
release_firmware_map_entry() should not call firmware_map_find_entry()
since releaed entry has been deleted from map_entries list. So the
patch delete firmware_map_find_entry() from releae_firmware_map_entry()
Shaohua Li [Mon, 29 Apr 2013 22:08:36 +0000 (15:08 -0700)]
mm: thp: add split tail pages to shrink page list in page reclaim
In page reclaim, huge page is split. split_huge_page() adds tail pages
to LRU list. Since we are reclaiming a huge page, it's better we
reclaim all subpages of the huge page instead of just the head page.
This patch adds split tail pages to shrink page list so the tail pages
can be reclaimed soon.
Before this patch, run a swap workload:
thp_fault_alloc 3492
thp_fault_fallback 608
thp_collapse_alloc 6
thp_collapse_alloc_failed 0
thp_split 916
With this patch:
thp_fault_alloc 4085
thp_fault_fallback 16
thp_collapse_alloc 90
thp_collapse_alloc_failed 0
thp_split 1272
mm: allow for outstanding swap writeback accounting
To prevent flooding the swap device with writebacks, frontswap backends
need to count and limit the number of outstanding writebacks. The
incrementing of the counter can be done before the call to
__swap_writepage(). However, the caller must receive a notification
when the writeback completes in order to decrement the counter.
To achieve this functionality, this patch modifies __swap_writepage() to
take the bio completion callback function as an argument.
end_swap_bio_write(), the normal bio completion function, is also made
non-static so that code doing the accounting can call it after the
accounting is done.
There should be no behavioural change to existing code.
mm: break up swap_writepage() for frontswap backends
swap_writepage() is currently where frontswap hooks into the swap write
path to capture pages with the frontswap_store() function. However, if
a frontswap backend wants to "resume" the writeback of a page to the
swap device, it can't call swap_writepage() as the page will simply
reenter the backend.
This patch separates swap_writepage() into a top and bottom half, the
bottom half named __swap_writepage() to allow a frontswap backend, like
zswap, to resume writeback beyond the frontswap_store() hook.
__add_to_swap_cache() is also made non-static so that the page for which
writeback is to be resumed can be added to the swap cache.
Fix a corner case for MAP_FIXED when requested mapping length is larger
than rlimit for virtual memory. In such case any overlapping mappings
are unmapped before we check for the limit and return ENOMEM.
The check is moved before the loop that unmaps overlapping parts of
existing mappings. When we are about to hit the limit (currently mapped
pages + len > limit) we scan for overlapping pages and check again
accounting for them.
This fixes situation when userspace program expects that the previous
mappings are preserved after the mmap() syscall has returned with error.
(POSIX clearly states that successfull mapping shall replace any
previous mappings.)
This corner case was found and can be tested with LTP testcase:
In this case the mmap, which is clearly over current limit, unmaps
dynamic libraries and the testcase segfaults right after returning into
userspace.
I've also looked at the second instance of the unmapping loop in the
do_brk(). The do_brk() is called from brk() syscall and from vm_brk().
The brk() syscall checks for overlapping mappings and bails out when
there are any (so it can't be triggered from the brk syscall). The
vm_brk() is called only from binmft handlers so it shouldn't be
triggered unless binmft handler created overlapping mappings.
Anton Vorontsov [Mon, 29 Apr 2013 22:08:31 +0000 (15:08 -0700)]
memcg: add memory.pressure_level events
With this patch userland applications that want to maintain the
interactivity/memory allocation cost can use the pressure level
notifications. The levels are defined like this:
The "low" level means that the system is reclaiming memory for new
allocations. Monitoring this reclaiming activity might be useful for
maintaining cache level. Upon notification, the program (typically
"Activity Manager") might analyze vmstat and act in advance (i.e.
prematurely shutdown unimportant services).
The "medium" level means that the system is experiencing medium memory
pressure, the system might be making swap, paging out active file
caches, etc. Upon this event applications may decide to further analyze
vmstat/zoneinfo/memcg or internal memory usage statistics and free any
resources that can be easily reconstructed or re-read from a disk.
The "critical" level means that the system is actively thrashing, it is
about to out of memory (OOM) or even the in-kernel OOM killer is on its
way to trigger. Applications should do whatever they can to help the
system. It might be too late to consult with vmstat or any other
statistics, so it's advisable to take an immediate action.
The events are propagated upward until the event is handled, i.e. the
events are not pass-through. Here is what this means: for example you
have three cgroups: A->B->C. Now you set up an event listener on
cgroups A, B and C, and suppose group C experiences some pressure. In
this situation, only group C will receive the notification, i.e. groups
A and B will not receive it. This is done to avoid excessive
"broadcasting" of messages, which disturbs the system and which is
especially bad if we are low on memory or thrashing. So, organize the
cgroups wisely, or propagate the events manually (or, ask us to
implement the pass-through events, explaining why would you need them.)
Performance wise, the memory pressure notifications feature itself is
lightweight and does not require much of bookkeeping, in contrast to the
rest of memcg features. Unfortunately, as of current memcg
implementation, pages accounting is an inseparable part and cannot be
turned off. The good news is that there are some efforts[1] to improve
the situation; plus, implementing the same, fully API-compatible[2]
interface for CONFIG_MEMCG=n case (e.g. embedded) is also a viable
option, so it will not require any changes on the userland side.
mm: madvise: complete input validation before taking lock
In madvise(), there doesn't seem to be any reason for taking the
¤t->mm->mmap_sem before start and len_in have been validated.
Incidentally, this removes the need for the out: label.
David Rientjes [Mon, 29 Apr 2013 22:08:22 +0000 (15:08 -0700)]
mm, hotplug: avoid compiling memory hotremove functions when disabled
__remove_pages() is only necessary for CONFIG_MEMORY_HOTREMOVE. PowerPC
pseries will return -EOPNOTSUPP if unsupported.
Adding an #ifdef causes several other functions it depends on to also
become unnecessary, which saves in .text when disabled (it's disabled in
most defconfigs besides powerpc, including x86). remove_memory_block()
becomes static since it is not referenced outside of
drivers/base/memory.c.
Build tested on x86 and powerpc with CONFIG_MEMORY_HOTREMOVE both enabled
and disabled.
mm: change __remove_pages() to call release_mem_region_adjustable()
Change __remove_pages() to call release_mem_region_adjustable(). This
allows a requested memory range to be released from the iomem_resource
table even if it does not match exactly to an resource entry but still
fits into. The resource entries initialized at bootup usually cover the
whole contiguous memory ranges and may not necessarily match with the
size of memory hot-delete requests.
If release_mem_region_adjustable() failed, __remove_pages() emits a
warning message and continues to proceed as it was the case with
release_mem_region(). release_mem_region(), which is defined to
__release_region(), emits a warning message and returns no error since a
void function.
Add release_mem_region_adjustable(), which releases a requested region
from a currently busy memory resource. This interface adjusts the
matched memory resource accordingly even if the requested region does
not match exactly but still fits into.
This new interface is intended for memory hot-delete. During bootup,
memory resources are inserted from the boot descriptor table, such as
EFI Memory Table and e820. Each memory resource entry usually covers
the whole contigous memory range. Memory hot-delete request, on the
other hand, may target to a particular range of memory resource, and its
size can be much smaller than the whole contiguous memory. Since the
existing release interfaces like __release_region() require a requested
region to be exactly matched to a resource entry, they do not allow a
partial resource to be released.
This new interface is restrictive (i.e. release under certain
conditions), which is consistent with other release interfaces,
__release_region() and __release_resource(). Additional release
conditions, such as an overlapping region to a resource entry, can be
supported after they are confirmed as valid cases.
There is no change to the existing interfaces since their restriction is
valid for I/O resources.
resource: add __adjust_resource() for internal use
Add __adjust_resource(), which is called by adjust_resource() internally
after the resource_lock is held. There is no interface change to
adjust_resource(). This change allows other functions to call
__adjust_resource() internally while the resource_lock is held.
Minchan Kim [Mon, 29 Apr 2013 22:08:15 +0000 (15:08 -0700)]
THP: fix comment about memory barrier
Currently the memory barrier in __do_huge_pmd_anonymous_page doesn't
work. Because lru_cache_add_lru uses pagevec so it could miss spinlock
easily so above rule was broken so user might see inconsistent data.
I was not first person who pointed out the problem. Mel and Peter
pointed out a few months ago and Peter pointed out further that even
spin_lock/unlock can't make sure of it:
At last, Hugh pointed out that even we don't need memory barrier in
there because __SetPageUpdate already have done it from Nick's commit 0ed361dec369 ("mm: fix PageUptodate data race") explicitly.
So this patch fixes comment on THP and adds same comment for
do_anonymous_page, too because everybody except Hugh was missing that.
It means we need a comment about that.
Andrew Shewmaker [Mon, 29 Apr 2013 22:08:12 +0000 (15:08 -0700)]
mm: reinititalise user and admin reserves if memory is added or removed
Alter the admin and user reserves of the previous patches in this series
when memory is added or removed.
If memory is added and the reserves have been eliminated or increased
above the default max, then we'll trust the admin.
If memory is removed and there isn't enough free memory, then we need to
reset the reserves.
Otherwise keep the reserve set by the admin.
The reserve reset code is the same as the reserve initialization code.
I tested hot addition and removal by triggering it via sysfs. The
reserves shrunk when they were set high and memory was removed. They
were reset higher when memory was added again.
projects / linux.git / log