[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH 2/6] block/nvme: Map doorbells pages write-only
From: |
Philippe Mathieu-Daudé |
Subject: |
Re: [PATCH 2/6] block/nvme: Map doorbells pages write-only |
Date: |
Tue, 22 Sep 2020 11:36:03 +0200 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.11.0 |
On 9/22/20 11:04 AM, Paolo Bonzini wrote:
> On 22/09/20 10:41, Philippe Mathieu-Daudé wrote:
>>> Besides looking more correct in access mode, is there any side effect
>>> of WO mapping?
>> TBH I don't have enough knowledge to answer this question.
>> I tested successfully on X86. I'm writing more tests.
>
> No problem with doing this, but PROT_WRITE does not work at all on x86.
> :) PROT_EXEC works if you have a machine with PKRU, but PROT_WRITE
> silently becomes PROT_READ|PROT_WRITE because the processor does not
> support it.
Ah this is why it works the same way in my testing.
I'll run tests on ARM.
Thanks,
Phil.
>
> Paolo
>
[PATCH 3/6] block/nvme: Reduce I/O registers scope, Philippe Mathieu-Daudé, 2020/09/21
[PATCH 4/6] block/nvme: Drop NVMeRegs structure, directly use NvmeBar, Philippe Mathieu-Daudé, 2020/09/21
[PATCH 5/6] block/nvme: Use register definitions from 'block/nvme.h', Philippe Mathieu-Daudé, 2020/09/21
[PATCH 6/6] block/nvme: Replace magic value by SCALE_MS definition, Philippe Mathieu-Daudé, 2020/09/21
Re: [PATCH 0/6] block/nvme: Map doorbells pages write-only, remove magic from nvme_init, no-reply, 2020/09/21
Re: [PATCH 0/6] block/nvme: Map doorbells pages write-only, remove magic from nvme_init, no-reply, 2020/09/21