Re: 2.6.8-rc2 crash(s)?

["Randy.Dunlap" <rddunlap@osdl.org>]

On Thu, 29 Jul 2004 18:22:47 -0400 Gene Heskett wrote:

| On Sunday 25 July 2004 02:09, Denis Vlasenko wrote:
| >On Sunday 25 July 2004 07:12, Gene Heskett wrote:
| >> >Not much... at least you can look up that EIP in System.map.
| >> >Also, do you really need all that sound stuff?
| >>
| >> It seems to all come in with the main driver for the ALC650.  And
| >> it works pretty good once i got it figured out.  Everything but
| >> the bt878 audio, which is so far down the s/s+n isn't more than 30
| >> db.
| >
| >I meant "do not use it and see whether that helps".
| >
| >> c0164376 isn't a label, but its in between these two in the
| >> System.map c0164340 t prune_dcache
| >> c0164500 T shrink_dcache_sb
| >>
| >> But thats all I can deduce from here.
| >
| >Of course, it points _inside_ prune_dcache(), not at the very
| >first instruction of it.
| >
| >Do:
| >
| >objdump -d <file containing prune_dcache>.o >file.objdump
| 
| This worked rather nicely and I have a rather large dcacheDOTo.txt 
| file now.
| 
| >and
| >make path/to/<file containing prune_dcache>.s
| 
| But this ones still being difficult.  Make does want to generate it.  
| At best it claims that dcache.o is uptodate. I don't figure one file 
| is worth much without the other, so whats wrong with my syntax?
| [root@coyote fs]# make dcache.c>.s
| [root@coyote fs]# less .s
| Which contains "make: Nothing to be done for `dcache.c'."

This should be (without < > brackets):

make fs/dcache.s

or whatever directory it (dcache.c) is in.

| The vanishing post contained my .config for a 2.6.7 kernel, which had 
| a log snip of a very similar Oops but without the total crash.  That 
| at least is a quite measureable improvement.
| 
| So obviously I'm not understanding your use of the <> arrows yet.

They just bracket metadata or descriptive terms.  Take out the
angle brackets when you substitute live words there.

| Now, since my last post, which so far in about 8 hours, has not come 
| back from the list, I've run memtest86-3.1a for 12 full passes thru 
| this gigabyte of ram with no errors reported.  I've also gone thru 
| the init.d directory shutting off things that it doesn't appear are 
| of any use to me.
| 
| And, I've got a tail running on the log in another window.
| 
| >and using resulting .objdump and .s files, find exact
| >instruction and C code line where it died.


--
~Randy