Re: [Kgdb-bugreport] Handling a fault (due to access by us) gracefully
Status: Beta
Brought to you by:
jwessel
From: Tom R. <tr...@ke...> - 2004-12-19 20:01:51
|
On Sun, Dec 19, 2004 at 10:38:34AM +0530, Amit S. Kale wrote: > On Saturday 18 Dec 2004 10:57 pm, Tom Rini wrote: > > On Sat, Dec 18, 2004 at 02:22:22PM +0530, Amit S. Kale wrote: > > > Tom, George, > > > > > > What's the limitation of the way it's currently implemented? > > > > It's not implemented, beyond a range check. This doesn't catch faults > > caused by things such as 'bt' from a breakpoint at say sys_sync > > (connect, break sys_sync, c, type sync after logged in, 'bt', boom). > > I believe this was fixed in a patch sent out by Mithlesh. > > If it isn't, could you send me a debug packet log for a reproduction case? This particular bug may have been by that patch. And after talking with Deepak about the ARM stuff, there's similar sparse warnings already in the tree. Monday I'll rework it slightly (I don't want to add back in kgdb_useraccess) and see if the 'bt' part is still bogus. After that, George can you give me some other sample commands that may still blow up kgdb ? Thanks. -- Tom Rini http://gate.crashing.org/~trini/ |