Re: [PATCH v2 5/9] nvram: add nrf51_soc flash read method

[Li Qiang]

Paolo Bonzini <pbonzini@redhat.com> 于2020年6月29日周一 下午11:32写道：
>
> On 29/06/20 13:55, P J P wrote:
> > |
> > | I prefer here just 'qemu_log_mask(LOG_UNIMP, "%s not implemented\n",
> > | __func__);' as other patches does.
> >
> > Earlier patch v1 did that. It was suggested to return ldl_le_p(), as that's 
> > a
> > valid return IIUC, instead of a zero(0), in case flash_read() is called.
> >
> > Thanks so much for the reviews. I'll send a revised series with due updates.
>
> I think abort() is preferable (while LOG_UNIMP is wrong as it implies
> there is something to do that QEMU is not doing).
>

Oh, here the UNIMP I understand as it will not be implemented, not the
thing 'should do but not do now'.

If. we use. abort(), the guest also. can trigger this abort(crash).
Though there is also other places we use this I think it is not very good.

In fact I would like to silent ignore(callback do nothing) if the
developer doesn't provide callback at the beginning.
But there are some cases we use LOG_UNIMP:
-->https://git.qemu.org/?p=qemu.git;a=commit;h=158b659451


Thanks,
Li Qiang


> Paolo
>