linux-toradex.git/lib/dma-debug.c, branch v4.4-rc6

dma-debug: Fix dma_debug_entry offset calculation

2015-12-16T19:24:26+00:00

dma-debug uses struct dma_debug_entry to keep track of dma coherent
memory allocation requests. The virtual address is converted into a pfn
and an offset. Previously, the offset was calculated using an incorrect
bit mask.  As a result, we saw incorrect error messages from dma-debug
like the following:

"DMA-API: exceeded 7 overlapping mappings of cacheline 0x03e00000"

Cacheline 0x03e00000 does not exist on our platform.

Cc: 
Fixes: 0abdd7a81b7e ("dma-debug: introduce debug_dma_assert_idle()")
Signed-off-by: Daniel Mentz 
Signed-off-by: Dan Williams

dma-debug: check nents in dma_sync_sg*

2015-11-07T01:50:42+00:00

Like dma_unmap_sg, dma_sync_sg* should be called with the original number
of entries passed to dma_map_sg, so do the same check in the sync path as
we do in the unmap path.

Signed-off-by: Robin Murphy 
Cc: Arnd Bergmann 
Cc: Marek Szyprowski 
Cc: Sumit Semwal 
Cc: Sakari Ailus 
Cc: Russell King 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

debugfs: Pass bool pointer to debugfs_create_bool()

2015-10-04T10:36:07+00:00

Its a bit odd that debugfs_create_bool() takes 'u32 *' as an argument,
when all it needs is a boolean pointer.

It would be better to update this API to make it accept 'bool *'
instead, as that will make it more consistent and often more convenient.
Over that bool takes just a byte.

That required updates to all user sites as well, in the same commit
updating the API. regmap core was also using
debugfs_{read|write}_file_bool(), directly and variable types were
updated for that to be bool as well.

Signed-off-by: Viresh Kumar 
Acked-by: Mark Brown 
Acked-by: Charles Keepax 
Signed-off-by: Greg Kroah-Hartman

dma-debug: skip debug_dma_assert_idle() when disabled

2015-07-17T23:39:53+00:00

If dma-debug is disabled due to a memory error, DMA unmaps do not affect
the dma_active_cacheline radix tree anymore, and debug_dma_assert_idle()
can print false warnings.

Disable debug_dma_assert_idle() when dma_debug_disabled() is true.

Signed-off-by: Haggai Eran 
Fixes: 0abdd7a81b7e ("dma-debug: introduce debug_dma_assert_idle()")
Cc: Dan Williams 
Cc: Joerg Roedel 
Cc: Vinod Koul 
Cc: Russell King 
Cc: James Bottomley 
Cc: Florian Fainelli 
Cc: Sebastian Ott 
Cc: Jiri Kosina 
Cc: Horia Geanta 
Cc: 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

lib/dma-debug: fix bucket_find_contain()

2015-04-17T13:03:54+00:00

bucket_find_contain() will search the bucket list for a dma_debug_entry.
When the entry isn't found it needs to search other buckets too, since
only the start address of a dma range is hashed (which might be in a
different bucket).

A copy of the dma_debug_entry is used to get the previous hash bucket
but when its list is searched the original dma_debug_entry is to be used
not its modified copy.

This fixes false "device driver tries to sync DMA memory it has not allocated"
warnings.

Signed-off-by: Sebastian Ott 
Cc: Florian Fainelli 
Cc: Horia Geanta 
Cc: Jiri Kosina 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

dma-debug: prevent early callers from crashing

2014-12-11T01:41:02+00:00

dma_debug_init() is called by architecture specific code at different
levels, but typically as a fs_initcall due to the debugfs initialization.
Some platforms may have early callers of the DMA-API, running prior to the
fs_initcall() level, which is not much of an issue unless
CONFIG_DMA_API_DEBUG is set.  When the DMA-API debugging facilities are
turned on a caller will go through:

debug_dma_map_{single,page}
  -> dma_mapping_error (inline function usually)
    -> debug_dma_mapping_error
      -> get_hash_bucket

Calling get_hash_bucket() returns a valid hash value since we hash on high
bits of the dma_addr cookie, but we will grab an unitialized spinlock,
which typically won't crash but produce a warning, the real crash will
however happen during the bucket list traversal because the list has not
been initialized yet.

An obvious solution is of course to move some of the offenders to run
after the fs_initcall level, but since this might not always be an option,
we add a flag "dma_debug_initialized" which is set to false by default,
and set to true once dma_debug_init() has had a chance to run.

The dma_debug_disabled() helper function previously introduced just needs
to check for dma_debug_initialized to allow the caller to proceed or not.

Signed-off-by: Florian Fainelli 
Cc: Dan Williams 
Cc: Jiri Kosina 
Cc: Horia Geanta 
Cc: Brian Norris 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

dma-debug: introduce dma_debug_disabled

2014-12-11T01:41:02+00:00

Add a helper function which returns whether the DMA debugging API is
disabled, right now we only check for global_disable, but in order to
accommodate early callers of the DMA-API, we will check for more
initialization flags in the next patch.

Signed-off-by: Florian Fainelli 
Cc: Dan Williams 
Cc: Jiri Kosina 
Cc: Horia Geanta 
Cc: Brian Norris 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

dma-debug: modify check_for_stack output

2014-09-02T12:12:13+00:00

s/fromstack/from stack

Signed-off-by: Horia Geanta 
Signed-off-by: Jiri Kosina

dma debug: account for cachelines and read-only mappings in overlap tracking

2014-03-04T15:55:47+00:00

While debug_dma_assert_idle() checks if a given *page* is actively
undergoing dma the valid granularity of a dma mapping is a *cacheline*.
Sander's testing shows that the warning message "DMA-API: exceeded 7
overlapping mappings of pfn..." is falsely triggering.  The test is
simply mapping multiple cachelines in a given page.

Ultimately we want overlap tracking to be valid as it is a real api
violation, so we need to track active mappings by cachelines.  Update
the active dma tracking to use the page-frame-relative cacheline of the
mapping as the key, and update debug_dma_assert_idle() to check for all
possible mapped cachelines for a given page.

However, the need to track active mappings is only relevant when the
dma-mapping is writable by the device.  In fact it is fairly standard
for read-only mappings to have hundreds or thousands of overlapping
mappings at once.  Limiting the overlap tracking to writable
(!DMA_TO_DEVICE) eliminates this class of false-positive overlap
reports.

Note, the radix gang lookup is sub-optimal.  It would be best if it
stopped fetching entries once the search passed a page boundary.
Nevertheless, this implementation does not perturb the original net_dma
failing case.  That is to say the extra overhead does not show up in
terms of making the failing case pass due to a timing change.

References:
  http://marc.info/?l=linux-netdev&m=139232263419315&w=2
  http://marc.info/?l=linux-netdev&m=139217088107122&w=2

Signed-off-by: Dan Williams 
Reported-by: Sander Eikelenboom 
Reported-by: Dave Jones 
Tested-by: Dave Jones 
Tested-by: Sander Eikelenboom 
Cc: Konrad Rzeszutek Wilk 
Cc: Francois Romieu 
Cc: Eric Dumazet 
Cc: Wei Liu 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds

dma-debug: fix overlap detection

2014-01-30T00:22:40+00:00

Commit 0abdd7a81b7e ("dma-debug: introduce debug_dma_assert_idle()") was
reworked to expand the overlap counter to the full range expressable by
3 tag bits, but it has a thinko in treating the overlap counter as a
pure reference count for the entry.

Instead of deleting when the reference-count drops to zero, we need to
delete when the overlap-count drops below zero.  Also, when detecting
overflow we can just test the overlap-count > MAX rather than applying
special meaning to 0.

Regression report available here:
http://marc.info/?l=linux-netdev&m=139073373932386&w=2

This patch, now tested on the original net_dma case, sees the expected
handful of reports before the eventual data corruption occurs.

Signed-off-by: Dan Williams 
Reported-by: Sander Eikelenboom 
Cc: Francois Romieu 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds