Re: [Qemu-devel] Should we introduce a TranslationRegion with its own codegen buffer?

[Alex Bennée]

Paolo Bonzini <address@hidden> writes:

> On 04/04/2016 13:24, Alex Bennée wrote:
>> FWIW I've dropped the patch that moves tb_find_fast out of tb_lock in
>> the upcoming base-patches-v2 series because it:
>>   - weirdly breaks the pxe tests
>
> I'm happy to take a look.

Be my guest:

https://github.com/stsquad/qemu/tree/mttcg/base-patches-v2-with-tb-fast-delock

It breaks on i386 and x86_64 softmmu targets with make check. I also got
a command line failing with a dump of the pxe bootfile and handbuilt seabios:

./x86_64-softmmu/qemu-system-x86_64 -display vnc=:0 -machine accel=tcg
-netdev user,id=net0,tftp=./,bootfile=tests/pxe-test-disk.raw -device
virtio-net-pci,netdev=net0,romfile=pc-bios/pxe-virtio.rom -chardev
stdio,id=seabios -device isa-debugcon,iobase=0x402,chardev=seabios -bios 
roms/seabios/out/bios.bin

>
>>   - is pretty ugly anyway
>
> What exactly makes it ugly?

Subjective I guess but the use of tb_lock_recursive() and
tb_lock_reset() seem not as clean as they could be. Is this stuff Fred
re-introduced?

>
> Paolo
>
>> This is obviously going to have an impact on performance which bought me
>> back to should we be thinking more holistically about how to approach
>> this problem.


--
Alex Bennée