linux-toradex.git/mm/dmapool.c, branch jb4.3_1.0.0-ga Linux kernel for Apalis and Colibri modules ENGR00266285-2 mm: dmapool: export for normal case 2013-06-19T10:13:58+00:00 Peter Chen peter.chen@freescale.com 2013-06-08T07:58:29+00:00 35b5874013dbc528c21c830043532e36c7bf84ca It is the same issue with ENGR00217721, and it is also needed for normal usb case. Signed-off-by: Peter Chen <peter.chen@freescale.com> 
It is the same issue with ENGR00217721, and it is also needed
for normal usb case.

Signed-off-by: Peter Chen <peter.chen@freescale.com>
ENGR00217721-4 implement dma_pool_alloc_nonbufferable interface 2012-07-25T05:11:33+00:00 Tony LIU junjie.liu@freescale.com 2012-07-20T09:53:55+00:00 651ce784926d727761ce510a0eb530c9e72e2793 mm core part - After USB driver prime a bulk transfer(whatever IN or OUT, take OUT for example) on ep1, only one dTD is primed, an USB Interrupt (bit 0 of USBSTS) will be issued, and find that endptcomplete register is 0x2 which means an OUT transfer on ep1 is completed, at this time the ep1 out queue head status is 0x1e18000, and next dtd pointer is 0x1 which means transfer is done and everything is OK, while the dTD token status is 0x2008080 which means this dTD is still active, not completed yet. - Audio SDMA and Ethernet have the similar issue - root cause is not found yet - work around: change the non-cacheable bufferable memory to non-cacheable non-bufferable memory to make this issue disappear. Signed-off-by: Tony LIU <junjie.liu@freescale.com> 
mm core part

- After USB driver prime a bulk transfer(whatever IN or OUT, take
  OUT for example) on ep1, only one dTD is primed, an USB Interrupt
  (bit 0 of USBSTS) will be issued, and find that endptcomplete
  register is 0x2 which means an OUT transfer on ep1 is completed,
  at this time the ep1 out queue head status is 0x1e18000, and next
  dtd pointer is 0x1 which means transfer is done and everything is
  OK, while the dTD token status is 0x2008080 which means this dTD
  is still active, not completed yet.
- Audio SDMA and Ethernet have the similar issue
- root cause is not found yet
- work around:
  change the non-cacheable bufferable memory to non-cacheable
  non-bufferable memory to make this issue disappear.

Signed-off-by: Tony LIU <junjie.liu@freescale.com>
mm/dmapool.c: use TASK_UNINTERRUPTIBLE in dma_pool_alloc() 2011-01-14T01:32:48+00:00 Andrew Morton akpm@linux-foundation.org 2011-01-13T23:47:25+00:00 684265d4a30f133162f06ddb2e5010608e60e4bb As it stands this code will degenerate into a busy-wait if the calling task has signal_pending(). Cc: Rolf Eike Beer <eike-kernel@sf-tec.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> 
As it stands this code will degenerate into a busy-wait if the calling task
has signal_pending().

Cc: Rolf Eike Beer <eike-kernel@sf-tec.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/dmapool.c: take lock only once in dma_pool_free() 2011-01-14T01:32:48+00:00 Rolf Eike Beer eike-kernel@sf-tec.de 2011-01-13T23:47:24+00:00 84bc227d7fde049a568cd58a5610613feedc0dff dma_pool_free() scans for the page to free in the pool list holding the pool lock. Then it releases the lock basically to acquire it immediately again. Modify the code to only take the lock once. This will do some additional loops and computations with the lock held in if memory debugging is activated. If it is not activated the only new operations with this lock is one if and one substraction. Signed-off-by: Rolf Eike Beer <eike-kernel@sf-tec.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> 
dma_pool_free() scans for the page to free in the pool list holding the
pool lock.  Then it releases the lock basically to acquire it immediately
again.  Modify the code to only take the lock once.

This will do some additional loops and computations with the lock held in
if memory debugging is activated.  If it is not activated the only new
operations with this lock is one if and one substraction.

Signed-off-by: Rolf Eike Beer <eike-kernel@sf-tec.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm: add a might_sleep_if() to dma_pool_alloc() 2010-10-26T23:52:08+00:00 Dima Zavin dima@android.com 2010-10-26T21:21:54+00:00 ea05c8444e451f1cfbf78c68733e717ad7b8602b Buggy drivers (e.g. fsl_udc) could call dma_pool_alloc from atomic context with GFP_KERNEL. In most instances, the first pool_alloc_page call would succeed and the sleeping functions would never be called. This allowed the buggy drivers to slip through the cracks. Add a might_sleep_if() checking for __GFP_WAIT in flags. Signed-off-by: Dima Zavin <dima@android.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> 
Buggy drivers (e.g.  fsl_udc) could call dma_pool_alloc from atomic
context with GFP_KERNEL.  In most instances, the first pool_alloc_page
call would succeed and the sleeping functions would never be called.  This
allowed the buggy drivers to slip through the cracks.

Add a might_sleep_if() checking for __GFP_WAIT in flags.

Signed-off-by: Dima Zavin <dima@android.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
dmapools: protect page_list walk in show_pools() 2009-07-01T01:56:00+00:00 Thomas Gleixner tglx@linutronix.de 2009-06-30T18:41:25+00:00 c49568235dd7b4a2ffad63aa950562f4ffb9455f show_pools() walks the page_list of a pool w/o protection against the list modifications in alloc/free. Take pool->lock to avoid stomping into nirvana. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Matthew Wilcox <willy@linux.intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> 
show_pools() walks the page_list of a pool w/o protection against the list
modifications in alloc/free.  Take pool->lock to avoid stomping into
nirvana.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
dmapool: enable debugging for CONFIG_SLUB_DEBUG_ON too 2008-04-28T15:58:20+00:00 Andi Kleen ak@suse.de 2008-04-28T09:12:37+00:00 b5ee5befa75e33e55d34584ad10286c5005cb1de Previously it was only enabled for CONFIG_DEBUG_SLAB. Not hooked into the slub runtime debug configuration, so you currently only get it with CONFIG_SLUB_DEBUG_ON, not plain CONFIG_SLUB_DEBUG Acked-by: Matthew Wilcox <willy@linux.intel.com> Signed-off-by: Andi Kleen <ak@suse.de> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> 
Previously it was only enabled for CONFIG_DEBUG_SLAB.

Not hooked into the slub runtime debug configuration, so you currently only
get it with CONFIG_SLUB_DEBUG_ON, not plain CONFIG_SLUB_DEBUG

Acked-by: Matthew Wilcox <willy@linux.intel.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
pool: Improve memory usage for devices which can't cross boundaries 2007-12-04T15:39:58+00:00 Matthew Wilcox matthew@wil.cx 2007-12-03T19:16:24+00:00 e34f44b3517fe545f7fd45a8c2f6ee1e5e4432d3 The previous implementation simply refused to allocate more than a boundary's worth of data from an entire page. Some users didn't know this, so specified things like SMP_CACHE_BYTES, not realising the horrible waste of memory that this was. It's fairly easy to correct this problem, just by ensuring we don't cross a boundary within a page. This even helps drivers like EHCI (which can't cross a 4k boundary) on machines with larger page sizes. Signed-off-by: Matthew Wilcox <willy@linux.intel.com> Acked-by: David S. Miller <davem@davemloft.net> 
The previous implementation simply refused to allocate more than a
boundary's worth of data from an entire page.  Some users didn't know
this, so specified things like SMP_CACHE_BYTES, not realising the
horrible waste of memory that this was.  It's fairly easy to correct
this problem, just by ensuring we don't cross a boundary within a page.
This even helps drivers like EHCI (which can't cross a 4k boundary)
on machines with larger page sizes.

Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
Acked-by: David S. Miller <davem@davemloft.net>
Change dmapool free block management 2007-12-04T15:39:57+00:00 Matthew Wilcox matthew@wil.cx 2007-12-03T19:08:28+00:00 a35a3455142976e3fffdf27027f3082cbaba6e8c Use a list of free blocks within a page instead of using a bitmap. Update documentation to reflect this. As well as being a slight reduction in memory allocation, locked ops and lines of code, it speeds up a transaction processing benchmark by 0.4%. Signed-off-by: Matthew Wilcox <willy@linux.intel.com> 
Use a list of free blocks within a page instead of using a bitmap.
Update documentation to reflect this.  As well as being a slight
reduction in memory allocation, locked ops and lines of code, it speeds
up a transaction processing benchmark by 0.4%.

Signed-off-by: Matthew Wilcox <willy@linux.intel.com>
dmapool: Tidy up includes and add comments 2007-12-04T15:39:57+00:00 Matthew Wilcox matthew@wil.cx 2007-12-03T17:16:57+00:00 6182a0943af2235756836ed7e021fa22b93ec68b We were missing a copyright statement and license, so add GPLv2, David Brownell's copyright and my copyright. The asm/io.h include was superfluous, but we were missing a few other necessary includes. Signed-off-by: Matthew Wilcox <willy@linux.intel.com> 
We were missing a copyright statement and license, so add GPLv2, David
Brownell's copyright and my copyright.

The asm/io.h include was superfluous, but we were missing a few other
necessary includes.

Signed-off-by: Matthew Wilcox <willy@linux.intel.com>