Re: [PATCH hurd] libports: use protected payloads to optimize the object lookup

[Samuel Thibault]

Justus Winter, le Wed 22 Apr 2015 01:58:33 +0200, a écrit :
> Quoting Samuel Thibault (2015-04-20 04:17:53)
> > Boot it, use all the default parameters, and just after partition
> > reloading from the kernel (which prints the new partition layout), the
> > system is completely frozen.
> 
> Please check whether or not the root filesystem in fact dies.
> /hurd/startup doesn't detect that :/

gnumach still shows the task, but it's inside exception catchers:

db> show task $task1
    TASK        THREADS
  1 ext2fs (f4d68e40): 16 threads:
              0 (f4d693c0) ..SO.F(walking_zombie)
              1 (f4d69140) .W.O.F(exception_raise_continue) 0
              2 (f4d69000) .W.O..(mach_msg_receive_continue) 0
              3 (f4db2dc8) .W.O..(mach_msg_continue) 0
              4 (f4db2c88) .W.O..(mach_msg_receive_continue) 0
              5 (f4db2b48) .W.O.F(mach_msg_continue) 0
              6 (f4db2a08) .W.O.F(exception_raise_continue) 0
              7 (f4db28c8) .W.O.F(exception_raise_continue) 0
              8 (f4db2288) .W.O..(mach_msg_receive_continue) 0
              9 (f4dc9c90) R.....
             10 (f4dc9a10) .W.O.F(exception_raise_continue) 0
             11 (f4dc9790) .W.O.F(exception_raise_continue) 0
             12 (f4e4fe00) R..O.F(swtch_pri_continue)
             13 (f4e4f680) .W.O.F(mach_msg_continue) 0
             14 (f4e64e08) R..O.F(swtch_pri_continue)
             15 (f4ebf568) ..SO..(thread_exception_return)
db> trace/tu $task1.1
Continuation exception_raise_continue
>>>>> user space <<<<<
0x807d61a(0,805de90,81ddff0,0,0)
Bad frame pointer: 0x1d4c0
db> trace/tu $task1.6
Continuation exception_raise_continue
>>>>> user space <<<<<
0x807d61a(0,805de90,81ddff0,0,0)
Bad frame pointer: 0x1d4c0
db> trace/tu $task1.7
Continuation exception_raise_continue
>>>>> user space <<<<<
0x807d61a(0,805de90,81ddff0,0,0)
Bad frame pointer: 0x1d4c0
db> trace/tu $task1.10
Continuation exception_raise_continue
>>>>> user space <<<<<
0x807d61a(0,805de90,81ddff0,0,0)
Bad frame pointer: 0x1d4c0
db> trace/tu $task1.11
Continuation exception_raise_continue
>>>>> user space <<<<<
0x807d61a(0,805de90,81ddff0,0,0)
Bad frame pointer: 0x1d4c0

$ addr2line -e ext2fs/ext2fs.static  0x807d61a
/usr/src/hurd-debian/build-deb/libports/../../libports/ports.h:268

I'm rebuilding an ext2fs.static with -fno-omit-frame-pointer to get more
backtrace.

Samuel