Bug 639824
| Summary: | dl2k driver tries to free an invalid DMA memory address | ||||||
|---|---|---|---|---|---|---|---|
| Product: | [Fedora] Fedora | Reporter: | Frantisek Hanzlik <franta> | ||||
| Component: | kernel | Assignee: | Stanislaw Gruszka <sgruszka> | ||||
| Status: | CLOSED UPSTREAM | QA Contact: | Fedora Extras Quality Assurance <extras-qa> | ||||
| Severity: | medium | Docs Contact: | |||||
| Priority: | low | ||||||
| Version: | 14 | CC: | dougsland, gansalmon, itamar, jonathan, kernel-maint, madhu.chinakonda, sgruszka | ||||
| Target Milestone: | --- | ||||||
| Target Release: | --- | ||||||
| Hardware: | i686 | ||||||
| OS: | Linux | ||||||
| Whiteboard: | |||||||
| Fixed In Version: | Doc Type: | Bug Fix | |||||
| Doc Text: | Story Points: | --- | |||||
| Clone Of: | Environment: | ||||||
| Last Closed: | 2011-02-05 20:48:02 UTC | Type: | --- | ||||
| Regression: | --- | Mount Type: | --- | ||||
| Documentation: | --- | CRM: | |||||
| Verified Versions: | Category: | --- | |||||
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||
| Cloudforms Team: | --- | Target Upstream Version: | |||||
| Embargoed: | |||||||
| Attachments: |
|
||||||
|
Description
Frantisek Hanzlik
2010-10-04 07:47:20 UTC
Created attachment 452534 [details]
f13-dl2k-nulify-fraginfo-after-unmap.patch
Proposed fix. Can you apply it, build and test kernel? If not I will prepare kernel rpm with patch.
Please try this kernel when it finish to build: http://koji.fedoraproject.org/koji/taskinfo?taskID=2530611 I've compiled dl2k.ko with Your patch for actual 2.6.35.6-39.fc14.i686.PAE kernel, but this computer will be accessible for me as late as end of this week. And it appears that Your kernel is 2.6.34.7-59.bz639824.fc13.i686, not 2.6.35/F14, and I w\don't want return to this version. There was probably some dl2k development lately, as this card works for me in FC5/2.6.17.axp and FC6/2.6.22.i686 distro/kernels, but not in FC10/2.6.27.i686 kernel, and I not sure when will work in 2.6.34.x kernels. I will report how it works at beginning of next week. (In reply to comment #3) > And it appears that Your kernel is 2.6.34.7-59.bz639824.fc13.i686, not > 2.6.35/F14, Oops, sorry I missed system version. If you are able to apply a patch that's great. I'm waiting for testing results. Hello Stanislaw, this server with PCI-X Fiber 1000BASE-SX D-Link Network Adapter DGE-550SX (in 32bit slot) and Your dl2k patch on 2.6.35.6-39.fc14.i686.PAE Fedora kernel now works for 3+ days without problems and without any oopses in /var/log/messages. As it works 24x7 with several tens (more likely hunderts) gigabytes every day throught this D-Link card, I thing this problem is away. One interesant thing which I cannot understand and You maybe should clarify that - folks who owned this server performed at 13-Oct kernel upgrade to version 2.6.35.6-39.fc14.i686.PAE. And although, with previous 2.6.35.4-28.fc14.i686.PAE kernel, above oopses occurs always after from several minutes to several few hours (but only once, as I stated above), with this new kernel problem not occured for all four days when was running. I not know for what kernel/driver version You made this patch, but it was probably little different from this 2.6.35.6-39.fc14.i686.PAE in Fedora (which I patched), as patch program found Your changes at different location - but was patching it: $ patch -p0 <f13-dl2k-nulify-fraginfo-after-unmap.patch patching file a/drivers/net/dl2k.c Hunk #1 succeeded at 1753 (offset -2 lines). Hunk #2 succeeded at 1761 (offset -2 lines). It seems as 2.6.35.6-39.fc14.i686.PAE without Your patch works fine too. Or maybe within these four days not occured some conditions which are needed for this oops? I think bug is still there. It should be 100% reproducible with ifconfig ethX down. Perhaps on older kernel (or user space) there was condition which make device do the close, what not happens anymore after update. Patch land in linus tree: http://git.kernel.org/linus/9eb710797a21fa4a9e09ae9c86c4b3ec9d291c2d > It seems as 2.6.35.6-39.fc14.i686.PAE without Your patch works fine too. If bug does not bother you, I will not bother myself for fedora backport :-) Since fix is available on 2.6.38 and never, I'm closing bug with upstream resolution. |