gdb removing breakpoints after first breakpoint is hit ??

gdb removing breakpoints after first breakpoint is hit ??

[Jerone Young]

Has anyone notice with x86 that after you hit your first breakpoint and
come back into gdb from kvm, gdb starts to sends commands to the qemu
gdb stub to serially remove all the break points you have specified?
Ultimately ending up in a state where you will have no breakpoints when
you go back into KVM. 

So you end up hitting your first breakpoint and then no others (since
gdb removed them). Could be my version of gdb .. but trying to quickly
see if anyone else has seen this.

Re: gdb removing breakpoints after first breakpoint is hit ??

[Jan Kiszka]

[-- Attachment #1: Type: text/plain, Size: 1096 bytes --]

Jerone Young wrote:
> Has anyone notice with x86 that after you hit your first breakpoint and
> come back into gdb from kvm, gdb starts to sends commands to the qemu
> gdb stub to serially remove all the break points you have specified?
> Ultimately ending up in a state where you will have no breakpoints when
> you go back into KVM. 
> 
> So you end up hitting your first breakpoint and then no others (since
> gdb removed them). Could be my version of gdb .. but trying to quickly
> see if anyone else has seen this.

Not really, but maybe some error made gdb remove them. You have a trace
of the remote session (set debug remote 1)? Which gdb version? Which kvm
version? Note that Avi just merged latest qemu SVN into userland which
also brings in a few new features for the gdbstub: detach and kill. The
latter is clearly visible (kvm quits), the latter removes all
breakpoints as well. But that will be visible in the session trace.

Ah, and another pitfall is trying to debug SMP guests: Only VCPU 0 will
report breakpoints to you (see related patch of mine).

Jan


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 254 bytes --]