Messages in this thread | | | Date | Sun, 27 Dec 1998 07:14:19 +0000 | From | Alex Stewart <> | Subject | Re: [2.1.132] Modularized sound oops, OPL3-SA1 |
| |
(I picked this up while browsing through a kernel archive for last week)
> With a modularized opl3sa1 sound card setup, it looks like soundcore.o's > MOD_USE_COUNT is never fully decremented, even when all it's child modules > are (auto or manually) unloaded, and the next attempt at autoloading the > sound modules causes an oops.
I just wanted to note that I've seen the same behavior with 2.1.131 and my (fixed) pas2 driver, though I haven't yet had it oops on me. The symptoms are as follows:
(using Kernel 2.1.131 (with linux-2.1.131-pas.patch posted to this list earlier) with modutils 2.1.121 on a Pentium 75 (32MB RAM, 1.5GB HD))
[root@mondas /]# insmod soundcore [root@mondas /]# lsmod Module Size Used by soundcore 1968 0 (unused) [root@mondas /]# insmod sound [root@mondas /]# lsmod Module Size Used by sound 57084 0 (unused) soundcore 1968 3 [sound] [root@mondas /]# insmod pas2 io=0x388 irq=11 dma=7 [root@mondas /]# lsmod Module Size Used by pas2 11376 0 (unused) sound 57084 0 [pas2] soundcore 1968 6 [sound] [root@mondas /]# rmmod pas2 [root@mondas /]# lsmod Module Size Used by sound 57084 0 soundcore 1968 6 [sound] [root@mondas /]# insmod pas2 io=0x388 irq=11 dma=7 [root@mondas /]# lsmod Module Size Used by pas2 11376 0 (unused) sound 57084 0 [pas2] soundcore 1968 9 [sound] [root@mondas /]# rmmod pas2 [root@mondas /]# lsmod Module Size Used by sound 57084 0 soundcore 1968 9 [sound] [root@mondas /]# rmmod sound [root@mondas /]# lsmod Module Size Used by soundcore 1968 6
(equivalent behavior when using 'modprobe pas2' or demand-loading/etc.)
I note that the soundcore use count goes up by 3 when the sound module is loaded, which gets decremented properly when it's unloaded, but when the pas2 module is loaded it gets incremented by the same amount (3) which doesn't ever get decremented. Could the loading of the pas2 module be erroneously causing the sound module's use of soundcore to be counted again?
I don't know a lot about this aspect of the kernel stuff (just recently started looking into module stuff) but one wonders if this might be a problem with three(plus)-module chains (most other things I've seen don't have dependencies deeper than 2 levels). I don't know off hand of any 4-deep chains I could try with my hardware or I would..
Anyway, just some speculations.. I'll post more if I come up with anything more revealing.
Please CC any relevant questions/comments to me directly, as I am not currently subscribed to the linux-kernel maillist.
-alex
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |