I finally tracked down the root cause of the hangs - I had applied the modsign patches from the Fedora kernel, and hit several bugs. 1) The hangs in modprobe were caused by a bug in RedHat code (a function returned a bum pointer instead of NULL, and things went pear-shaped). I've reported this as bug 171835 in RedHat's bugzilla system. Redhat problem, not yours or lkml's. 2) It turns out that the issue in 171835 only bites if other things have already gone bad. It turns out that a patch in -rc5-mm1 causes the indigestion. clean-crypto-sha1c-up-a-bit.patch gives the modsign patches indigestion. I'm not sure why yet, or who's code is busticated. All I've verified for sure is that modsign allocates a 20-byte buffer, and with the cleanup patch applied, I'm seeing different values in buffer[12..19]. It's 4AM, and I'm not going to chase this any further until evening. *yawn* ;)
Attachment:
pgp502Exy2CQT.pgp
Description: PGP signature
- References:
- 2.6.14-rc5-mm1
- From: Andrew Morton <[email protected]>
- intel-agp and yenta-socket issues (was Re: 2.6.14-rc5-mm1
- From: [email protected]
- Re: intel-agp and yenta-socket issues (was Re: 2.6.14-rc5-mm1
- From: Andrew Morton <[email protected]>
- Re: intel-agp and yenta-socket issues (was Re: 2.6.14-rc5-mm1
- From: [email protected]
- 2.6.14-rc5-mm1
- Prev by Date: Re: [PATCH 2/8] VFS: per inode statfs (core)
- Next by Date: Re: dumb muliport serial cards not supported in 2.6.13.4 ???
- Previous by thread: Re: intel-agp and yenta-socket issues (was Re: 2.6.14-rc5-mm1
- Next by thread: Re: 2.6.14-rc5-mm1
- Index(es):