Re: [Qemu-devel] [RFC v3] monitor: add memory search commands s, sp

[Claudio Fontana]

On 21 April 2015 at 16:47, Paolo Bonzini <address@hidden> wrote:
>
>
> On 21/04/2015 16:35, Luiz Capitulino wrote:
>> > gladly, I am definitely looking for Review, first I wanted to check if the 
>> > idea itself was acceptable.
>>
>> I honestly don't have an opinion on this, which means I'm willing
>> to apply if there are no objections.
>>
>> On the other hand, isn't it possible to just use existing commands?
>> Like, why can't you just dump this memory to disk and search it
>> with strings or another tool?
>
> I don't think that can be a serious alternative.  The same would apply
> to x and xp, but still, they are incredibly useful.  Now I must say I
> don't know of a use case for s or sp, but I don't think what you
> suggested is feasible.
>
> Paolo

I must admit that the use case is a bit overlapping with gdb's
find-in-memory feature
(does it work with qemu's gdbserver?) which I however never use
because I tend to do everything from qemu.

The same could be said for the x and xp commands though, and in that
regard I think s and sp are a natural complement to them.

I have been using s and sp when dealing with code whose runtime
behavior is difficult to follow, to verify that the results of a
complex operation ended up at the right address.

For example, to find out where an ELF header landed in memory, and
verify that the loading address is correct,
looking for the TLS thread control block data structures, and so on.

If this is acceptable I will resubmit with a NULL initialization to
get rid of that spurious warning.

Thanks,

Claudio