RFR: 8342607: Enhance register printing on x86_64 platforms [v4]

[Martin Doerr]

On Mon, 28 Oct 2024 13:03:42 GMT, Thomas Stuefe <stuefe at openjdk.org> wrote:

>> Martin Doerr has updated the pull request incrementally with one additional commit since the last revision:
>> 
>>   Check uc->uc_mcontext.fpregs sanity.
>
> src/hotspot/os_cpu/linux_x86/os_linux_x86.cpp line 551:
> 
>> 549:   st->cr();
>> 550:   st->cr();
>> 551:   size_t fpregs_offset = pointer_delta(uc->uc_mcontext.fpregs, uc, 1);
> 
> Could the register substructure live outside uc on x64? If so, it may be safer to 
> Suggestion:
> 
>   size_t fpregs_offset = (uc->uc_mcontext.fpregs >= uc) ? pointer_delta(uc->uc_mcontext.fpregs, uc, 1) : 0;
> 
> or similar, since the register substructure may precede uc which would make pointer_delta assert.

I think using 0 would require more changes to avoid accessing uc+0 which would be wrong. Richard and I already discussed about this above. We think it's acceptable. Some other projects claim that FP register substructure is inside the uc: https://github.com/mono/mono/blob/0f53e9e151d92944cacab3e24ac359410c606df6/mono/utils/mono-sigcontext.h#L263
We could also check the kernel code which writes it.

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/21615#discussion_r1819079823