Søren Schmidt wrote: >> On Fri, Jul 02, 2004 at 10:58:59AM -0700, Nate Lawson wrote: >>> That's because the bug is still there. Stick a call to "if (addr == >>> 0xc1740a00) backtrace();" in kern_malloc and you'll see the original >>> allocator was ATA. The problem is with ATA's handling of drives >>> where you >>> get the message "identify timed out". To work around, boot without your >>> cdrom drive in the drive bay. This bug is extremely annoying and has >>> forced me to install via floppy in the past. >>> >>> The reason you don't see this in a custom kernel is the check for memory >>> use after free happens only with options INVARIANTS. The bug is still >>> there, just ignored. >>> >>> I may get around to tracking this down sometime in the future but don't >>> count on it. > > Yeps, some good precise debugging of this problem would be appreciated, > I can't seem to reproduce it here no matter what I try :( Just to close off this thread, I found the problems and posted a patch including debugging info to -current. -NateReceived on Fri Jul 30 2004 - 22:08:44 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:04 UTC