qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH] target/riscv: Fix satp write


From: LIU Zhiwei
Subject: Re: [PATCH] target/riscv: Fix satp write
Date: Mon, 6 Sep 2021 13:31:26 +0800
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101 Thunderbird/78.13.0


On 2021/9/6 上午11:26, Bin Meng wrote:
On Mon, Sep 6, 2021 at 11:23 AM LIU Zhiwei <zhiwei_liu@c-sky.com> wrote:

On 2021/9/2 上午10:47, Bin Meng wrote:
On Thu, Sep 2, 2021 at 10:44 AM LIU Zhiwei <zhiwei_liu@c-sky.com> wrote:
On 2021/9/2 上午9:59, Bin Meng wrote:
On Thu, Sep 2, 2021 at 9:02 AM LIU Zhiwei <zhiwei_liu@c-sky.com> wrote:
On 2021/9/1 下午9:05, Bin Meng wrote:
On Wed, Sep 1, 2021 at 8:51 PM LIU Zhiwei <zhiwei_liu@c-sky.com> wrote:
These variables should be target_ulong. If truncated to int,
the bool conditions they indicate will be wrong.

As satp is very important for Linux, this bug almost fails every boot.
Could you please describe which Linux configuration is broken?
I use the image from:

https://gitlab.com/c-sky/buildroot/-/jobs/1251564514/artifacts/browse/output/images/

     I have
a 64-bit 5.10 kernel and it boots fine.
The login is mostly OK for me. But the busybox can't run properly.
Which kernel version is this?
5.10.4
Could you please investigate and
indicate in the commit message?

I just tested current qemu-system-riscv64 can boot to Ubuntu 20.04
distro user space.
Very strange.  This will cause tlb_flush can't be called in this function.

Did your kernel enable asid?
Yes. Is it matter?
Not sure, the tbl_flush is on the ASID path. I suspect the kernel we
(Alistair and me) tested did not enable ASID.
In my opinion, if the ASID is open, we should not flush tlb when ASID changes in most cases.
If ASID is not open.
Regards,
Bin



reply via email to

[Prev in Thread] Current Thread [Next in Thread]