Skip to contentSkip to conversation

dma-debug fixes for 2.6.31-rc1

GIT PULL New Joerg Roedel 2 participants 4 messages Jul 3, 2009 Library code

lore
Share & tools
lore
#0 lore raw

Hi Ingo,

The following changes since commit 7c5371c403abb29f01bc6cff6c5096abdf2dc524:
Yinghai Lu (1):
x86: add boundary check for 32bit res before expand e820 resource to alignment

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/joro/linux-2.6-iommu.git dma-debug/fixes

Ingo Molnar (1):
dma-debug: Put all hash-chain locks into the same lock class

Joerg Roedel (1):
dma-debug: fix off-by-one error in overlap function

 lib/dma-debug.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

These patches make dma-debug work better with lockdep enabled and fix a corner case in the overlap function were one region ends exactly where the other one begins. Please pull.

Thanks,

Joerg

#1 lore raw

PATCH 1/2 dma-debug: fix off-by-one error in overlap function

This patch fixes a bug in the overlap function which returned true if one region ends exactly before the second region begins. This is no overlap but the function returned true in that case.

---
 lib/dma-debug.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
lib/dma-debug.c +1 -1
index 3b93129..a9b6b5c 100644
@@ -862,7 +862,7 @@ static inline bool overlap(void *addr, u64 size, void *start, void *end)
 
 	return ((addr >= start && addr < end) ||
 		(addr2 >= start && addr2 < end) ||
-		((addr < start) && (addr2 >= end)));
+		((addr < start) && (addr2 > end)));
 }
 
 static void check_for_illegal_area(struct device *dev, void *addr, u64 size)
Signature · 1 line
1.6.3.3
#2 lore raw

PATCH 2/2 dma-debug: Put all hash-chain locks into the same lock class

In reply to Joerg Roedel

From: Ingo Molnar <[email protected]>

Alan Cox reported that lockdep runs out of its stack-trace entries with certain configs:

 BUG: MAX_STACK_TRACE_ENTRIES too low

This happens because there are 1024 hash buckets, each with a separate lock. Lockdep puts each lock into a separate lock class and tracks them independently.

But in reality we never take more than one of the buckets, so they really belong into a single lock-class. Annotate the has bucket lock init accordingly.

[ Impact: reduce the lockdep footprint of dma-debug ]

---
 lib/dma-debug.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
lib/dma-debug.c +1 -1
index a9b6b5c..c9187fe 100644
@@ -716,7 +716,7 @@ void dma_debug_init(u32 num_entries)
 
 	for (i = 0; i < HASH_SIZE; ++i) {
 		INIT_LIST_HEAD(&dma_entry_hash[i].list);
-		dma_entry_hash[i].lock = SPIN_LOCK_UNLOCKED;
+		spin_lock_init(&dma_entry_hash[i].lock);
 	}
 
 	if (dma_debug_fs_init() != 0) {
Signature · 1 line
1.6.3.3
#3 lore raw
In reply to Joerg Roedel
Quoted 16 lines · Joerg Roedel

Hi Ingo,

The following changes since commit 7c5371c403abb29f01bc6cff6c5096abdf2dc524:
Yinghai Lu (1):
x86: add boundary check for 32bit res before expand e820 resource to alignment

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/joro/linux-2.6-iommu.git dma-debug/fixes

Ingo Molnar (1):
dma-debug: Put all hash-chain locks into the same lock class

Joerg Roedel (1):
dma-debug: fix off-by-one error in overlap function

Quoted 6 lines
 lib/dma-debug.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

These patches make dma-debug work better with lockdep enabled and fix a corner case in the overlap function were one region ends exactly where the other one begins. Please pull.

Pulled, thanks Joerg!

Ingo