Several weeks ago Nicolas Pitre and Russell King exchanged some emails
(on this list) regarding a possible processor or kernel bug. I've
encountered quite a few "memory violation" errors that might be of the
same flavor as those that Nicolas cited.
I haven't been able to investigate the problems very much because when
I run gdb on the core file the call stack is generally corrupt,
i.e. it looks like:
Core was generated by `/home/hughm/prog'.
Program terminated with signal 11, Segmentation fault.
#0 0x0 in ?? ()
When the program seg faults, this information gets logged to
/var/log/messages :
kernel: prog: memory violation at pc=0x0227152c, lr=0xbfffd138
kernel: (bad address=0x0322119c, code 3)
kernel:
kernel: pc :[0227152c] lr :[bfffd138]
kernel: sp : bfffd128 ip : bfffd128 fp : bfffd180
kernel: r10: 027037b0 r9 : 02763244 r8 : 40023f7d
kernel: r7 : 000656c3 r6 : 00000006 r5 : 02e90180 r4 : 0272cbd0
kernel: r3 : 000ff150 r2 : 02d45498 r1 : 02eff150 r0 : 02e24c5c
kernel: Flags: nzCv IRQs on FIQs on Mode USER_32
kernel: Function entered at [<02271404>] from [<02268a2c>]
kernel: Function entered at [<0226893c>] from [<0219049c>]
kernel: Function entered at [<02190354>] from [<026044d4>]
kernel: Function entered at [<02603da8>] from [<02096ac8>]
kernel: Function entered at [<02096aa4>] from [<0206c9f0>]
kernel: Function entered at [<0206c928>] from [<0201c4fc>]
kernel: Function entered at [<0201c388>] from [<0211cd30>]
kernel: Function entered at [<0211c1e0>] from [<0211bb00>]
kernel: Function entered at [<02119bc4>] from [<0211ee84>]
kernel: Function entered at [<0211d904>] from [<0211f10c>]
kernel: Function entered at [<0211ef44>] from [<02032e24>]
kernel: Function entered at [<0203289c>] from [<0261b5fc>]
kernel: Function entered at [<0261b53c>] from [<02000104>]
Has any further progress been made in isolating and resolving this
problem? Is there any easy way for me to check if I am having the
same problem that Nicolas has had? Have any patches been accepted to
resolve this? (Nicolas referred to a "hack" to the kernel in his
original message. Is this something that I should try?) Has a zImage
with this hack in it been made available?
This on a netwinder which is currently running a 2.0.31 kernel.
Thanks in advance for your help,
Hugh
unsubscribe: body of `unsubscribe linux-arm' to [EMAIL PROTECTED]
