From: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
To: Alexei Filippov <alexei.filippov@syntacore.com>,
palmer@dabbelt.com, alistair.francis@wdc.com,
bin.meng@windriver.com, liwei1518@gmail.com,
zhiwei_liu@linux.alibaba.com
Cc: qemu-riscv@nongnu.org, qemu-devel@nongnu.org
Subject: Re: [PATCH] target/riscv/cpu_helper.c: fix wrong exception raise
Date: Tue, 9 Apr 2024 09:11:30 -0300 [thread overview]
Message-ID: <e68e899e-444e-4f8b-9b76-9c178516330d@ventanamicro.com> (raw)
In-Reply-To: <20240329134527.1570936-1-alexei.filippov@syntacore.com>
On 3/29/24 10:45, Alexei Filippov wrote:
> Successed two stage translation, but failed pmp check can cause guest
> page fault instead of regular page fault.
>
> In case of execution ld instuction in VS mode we can
> face situation when two stages of translation was passed successfully,
> and if PMP check was failed first_stage_error variable going to be
> setted to false and raise_mmu_exception will raise
> RISCV_EXCP_LOAD_GUEST_ACCESS_FAULT(scause == 21) instead of
> RISCV_EXCP_LOAD_ACCESS_FAULT(scause == 5) and this is wrong, according
> to RISCV privileged spec sect. 3.7: Attempting to execute a load or
> load-reserved instruction which accesses a physical address within a
> PMP region without read permissions raises a load access-fault
> exception.
>
> Signed-off-by: Alexei Filippov <alexei.filippov@syntacore.com>
> ---
> target/riscv/cpu_helper.c | 4 +---
> 1 file changed, 1 insertion(+), 3 deletions(-)
>
> diff --git a/target/riscv/cpu_helper.c b/target/riscv/cpu_helper.c
> index bc70ab5abc..eaef1c2c62 100644
> --- a/target/riscv/cpu_helper.c
> +++ b/target/riscv/cpu_helper.c
> @@ -1408,9 +1408,7 @@ bool riscv_cpu_tlb_fill(CPUState *cs, vaddr address, int size,
> __func__, pa, ret, prot_pmp, tlb_size);
>
> prot &= prot_pmp;
> - }
> -
> - if (ret != TRANSLATE_SUCCESS) {
> + } else {
This solves the issue but we're leaving the 'first_stage_error' flag in an inconsistent
state - the call for get_physical_address_pmp() inside "if (ret == TRANSLATE_SUCCESS)" is
both a PMP error and also a non-first stage error, but now we're leaving it to 'true'.
Raise raise_mmu_exception() will give the expected PMP fault for the wrong reasons, since
it thinks that it's a first-stage fault when it's not. This will work now, but any
future change to raise_mmu_exception can break this logic.
I have an internal (not yet sent to review) fix for the same problem. In my case I was
supposed to have an INST_ACCESS_FAULT (1) and was getting an INST_GUEST_PAGE_FAULT (20).
I'll see if I can send it ASAP so you can have a look and see if it's also good for your
problem.
Thanks,
Daniel
> /*
> * Guest physical address translation failed, this is a HS
> * level exception
prev parent reply other threads:[~2024-04-09 12:11 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-03-29 13:45 [PATCH] target/riscv/cpu_helper.c: fix wrong exception raise Alexei Filippov
2024-04-09 12:11 ` Daniel Henrique Barboza [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=e68e899e-444e-4f8b-9b76-9c178516330d@ventanamicro.com \
--to=dbarboza@ventanamicro.com \
--cc=alexei.filippov@syntacore.com \
--cc=alistair.francis@wdc.com \
--cc=bin.meng@windriver.com \
--cc=liwei1518@gmail.com \
--cc=palmer@dabbelt.com \
--cc=qemu-devel@nongnu.org \
--cc=qemu-riscv@nongnu.org \
--cc=zhiwei_liu@linux.alibaba.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).