bugs on 2.6.29-rc2

[GoogleCodeExporter]

What steps will reproduce the problem?
1. trigger any swap activation (ie. allocate 1GB of memmory)

What is the expected output? What do you see instead?
Swap should work, and compress pages. Kernel bug. System after while is 
useless completly.

What version of the product are you using? On what operating system?
1. Linux kernel 2.6.29-rc2 (2.6.29-rc2-00013-gf3b8436 exactly)
2. compcache r262

Please provide any additional information below.

I was using memtester which do mlock, but any swap activation triggers 
this bug.

Original issue reported on code.google.com by witold.b...@gmail.com on 18 Jan 2009 at 6:25

Attachments:

• dmesg.txt

[GoogleCodeExporter]

This bug will only happen on 32-bit systems with "highmem". Its happening since 
we
kumap_atomic() *after* spin_unlock(pool->lock).

I found another problem too: in some cases we try to kmap_atomic() more than 1 
page
which will surely cause crash.

I hope to resolve these soon. Thanks for reporting this issue.

Original comment by nitingupta910@gmail.com on 21 Jan 2009 at 11:38

• Changed state: Started
• Added labels: ****
• Removed labels: ****

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

Try r268. Please let me know your test results.

Original comment by nitingupta910@gmail.com on 21 Jan 2009 at 1:01

• Added labels: ****
• Removed labels: ****

[GoogleCodeExporter]

Fix is in r268 (compcache-0.5.1). Please reopen if you still hit this issue.

Original comment by nitingupta910@gmail.com on 22 Jan 2009 at 5:42

• Changed state: Fixed
• Added labels: ****
• Removed labels: ****

[GoogleCodeExporter]

This fix removed problem. I installed r268 on 3 heavly loaded machines and 
everything works stable.

Thanks.

Original comment by witold.b...@gmail.com on 30 Jan 2009 at 5:01

• Added labels: ****
• Removed labels: ****