Bug 15638
| Summary: | Oops when loading IPMI | ||
|---|---|---|---|
| Product: | Platform Specific/Hardware | Reporter: | Xavier Bestel (xavier.bestel) |
| Component: | x86-64 | Assignee: | Corey Minyard (minyard) |
| Status: | RESOLVED OBSOLETE | ||
| Severity: | normal | CC: | akpm, alan, ian |
| Priority: | P1 | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Kernel Version: | 2.6.33-2-amd64 (stock Debian experimental) | Subsystem: | |
| Regression: | No | Bisected commit-id: | |
|
Description
Xavier Bestel
2010-03-27 14:16:54 UTC
Did this system work on a previous kernel version? If so, which one? Looking through the code, The only way I can see this happening was if there was another driver already installed with the same name and it did not have a proper drv->p->kobj. But that doesn't make any sense. No, it never worked before. But this system is quite young and was only tried with 2.6.32 and 2.6.33. I'm sorry I must confess my ignorance, but how could a different driver with the same name be loaded ? The package doesn't seem to contain that kind of thing. [xav@awak:~]$ find /lib/modules/2.6.33-2-amd64/ -name '*ipmi*' /lib/modules/2.6.33-2-amd64/kernel/drivers/char/ipmi /lib/modules/2.6.33-2-amd64/kernel/drivers/char/ipmi/ipmi_msghandler.ko /lib/modules/2.6.33-2-amd64/kernel/drivers/char/ipmi/ipmi_watchdog.ko /lib/modules/2.6.33-2-amd64/kernel/drivers/char/ipmi/ipmi_si.ko /lib/modules/2.6.33-2-amd64/kernel/drivers/char/ipmi/ipmi_poweroff.ko /lib/modules/2.6.33-2-amd64/kernel/drivers/char/ipmi/ipmi_devintf.ko So it fails with 2.6.32, also? The traceback you have shows that some patches have been backported from 2.6.34, I believe, though that's sometimes hard to tell. Looking at this, it is very strange, because this is occurring far before any device work occurs, at least according to the backtrace. This occurs while the driver is registering itself with the device model. Looking at the 2.6.34 code and backtrace and guessing a little, the only thing I could find to match the backtrace was if the driver name was already registered and the already registered driver was missing something in its data structure. I'm wondering if this debian release has something incorrectly backported. I'll try to pull that code and see. I see this has been linked from http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=571980 . The 326m will probably be installed in about a month (though perhaps without ipmi!) at which point I will be unable to test anything with it. Assuming this is the same issue, I could test alternate kernels if you could point me at any you want testing. Would simply isolating the kernel that introduces the issue help? |