Messages in this thread | | | Date | Fri, 07 May 1999 17:24:12 +0200 | From | Manfred Spraul <> | Subject | Re: Oops assist... [Solution + Patch] |
| |
Philip Gladstone wrote: > I'll bet that having 220 bytes of Oops is a lot better > than nothing - note that you only want to save the > binary data, not the text.
I thought about extending the current Oops report: * include the complete stack dump, including current (i.e. 8 kB on i386, 16 kB otherwise) ---> ability to reconstruct all local variables, in the complete call chain. ability to find buffer overruns on stack etc. * test all registers if they point to valid memory, and dump a few hundert bytes if the memory is valid. --> ability to analyze the structure where the bad pointer came from.
The overall dump would be 10-40 kB binary data. I know that this would be overkill for most oops'es, but disk space is cheap, and ksymoops could throw away most data unless e.g '--realy-verbose' is set.
Note: I usually use a kernel debugger for finding bugs during development. If I analyze an oops than the oops is from a linux end user - the fewer times I must ask for more details the better.
Regards, Manfred
- 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/
| |