From: Peter Maydell <peter.maydell@linaro.org>
To: Richard Henderson <richard.henderson@linaro.org>
Cc: "Edgar E. Iglesias" <edgar.iglesias@gmail.com>,
QEMU Developers <qemu-devel@nongnu.org>,
Laurent Vivier <laurent@vivier.eu>
Subject: Re: [PATCH v2 21/30] linux-user/microblaze: Fix SIGFPE si_codes
Date: Tue, 24 Aug 2021 17:55:49 +0100 [thread overview]
Message-ID: <CAFEAcA9FiqbvCh1xZWdknEJV=jnoxgZqz=ky-A7zXMM+Re74AQ@mail.gmail.com> (raw)
In-Reply-To: <20210822035537.283193-22-richard.henderson@linaro.org>
On Sun, 22 Aug 2021 at 04:55, Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Fix a typo for ESR_EC_DIVZERO, which is integral not floating-point.
> Fix the if ladder for decoding floating-point exceptions.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
> linux-user/microblaze/cpu_loop.c | 20 +++++++++++++++-----
> 1 file changed, 15 insertions(+), 5 deletions(-)
>
> diff --git a/linux-user/microblaze/cpu_loop.c b/linux-user/microblaze/cpu_loop.c
> index 9e07e52573..4a75c853b2 100644
> --- a/linux-user/microblaze/cpu_loop.c
> +++ b/linux-user/microblaze/cpu_loop.c
> @@ -81,15 +81,25 @@ void cpu_loop(CPUMBState *env)
> env->iflags &= ~(IMM_FLAG | D_FLAG);
> switch (env->esr & 31) {
> case ESR_EC_DIVZERO:
> - si_code = TARGET_FPE_FLTDIV;
> + si_code = TARGET_FPE_INTDIV;
> break;
> case ESR_EC_FPU:
> - si_code = 0;
> - if (env->fsr & FSR_IO) {
> + /*
> + * Note that the kernel passes along fsr as si_code
> + * if there's no recognized bit set. Possibly this
> + * implies that si_code is 0, but follow the structure.
> + */
In theory it should: the Microblaze processor reference guide
https://www.xilinx.com/support/documentation/sw_manuals/mb_ref_guide.pdf
defines only 5 bits in the FSR, all of which we look at here.
However our implementation provides two loopholes by which a
high bit might get set:
* our implementation of MTS rfsr, rX doesn't prevent high bits
being set by the guest
* our implementation of gdbstub writes to fsr doesn't prevent
high bits being set by the guest
I don't know whether the real h/w makes the reserved FSR high
bits RAZ/WI or not; the spec doesn't say either way.
> + si_code = env->fsr;
> + if (si_code & FSR_IO) {
> si_code = TARGET_FPE_FLTINV;
> - }
> - if (env->fsr & FSR_DZ) {
> + } else if (si_code & FSR_OF) {
> + si_code = TARGET_FPE_FLTOVF;
> + } else if (si_code & FSR_UF) {
> + si_code = TARGET_FPE_FLTUND;
> + } else if (si_code & FSR_DZ) {
> si_code = TARGET_FPE_FLTDIV;
> + } else if (si_code & FSR_DO) {
> + si_code = TARGET_FPE_FLTRES;
> }
> break;
> default:
Side note: our implementation will never set FSR_DO; we don't
implement the denormal number handling the FPU does, where:
* operations on input denormals return a QNaN and set FSR.DO
* output denormals are flushed to + or - zero, setting FSR.UF
Anyway,
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
thanks
-- PMM
next prev parent reply other threads:[~2021-08-24 16:57 UTC|newest]
Thread overview: 60+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-08-22 3:55 [PATCH v2 00/30] linux-user: Clean up siginfo_t handling Richard Henderson
2021-08-22 3:55 ` [PATCH v2 01/30] linux-user/aarch64: Set siginfo_t addr field for SIGTRAP signals Richard Henderson
2021-08-22 3:55 ` [PATCH v2 02/30] linux-user/arm: " Richard Henderson
2021-08-22 3:55 ` [PATCH v2 03/30] linux-user/arm: Use force_sig() to deliver fpa11 emulation SIGFPE Richard Henderson
2021-08-22 3:55 ` [PATCH v2 04/30] linux-user: Zero out target_siginfo_t in force_sig() Richard Henderson
2021-08-24 16:32 ` Philippe Mathieu-Daudé
2021-08-22 3:55 ` [PATCH v2 05/30] linux-user: Provide new force_sig_fault() function Richard Henderson
2021-08-24 16:36 ` Philippe Mathieu-Daudé
2021-08-22 3:55 ` [PATCH v2 06/30] linux-user: Provide new force_sigsegv_for_addr() function Richard Henderson
2021-08-24 16:20 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 07/30] linux-user/arm: Use force_sig_fault() Richard Henderson
2021-08-24 16:38 ` Philippe Mathieu-Daudé
2021-08-22 3:55 ` [PATCH v2 08/30] linux-user/aarch64: " Richard Henderson
2021-08-24 16:39 ` Philippe Mathieu-Daudé
2021-08-22 3:55 ` [PATCH v2 09/30] linux-user/alpha: Set TRAP_UNK for bugchk and unknown gentrap Richard Henderson
2021-08-24 16:22 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 10/30] linux-user/alpha: Set FPE_FLTUNK for gentrap ROPRAND Richard Henderson
2021-08-24 16:22 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 11/30] linux-user/alpha: Use force_sig_fault, force_sigsegv_code Richard Henderson
2021-08-24 16:24 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 12/30] linux-user/cris: " Richard Henderson
2021-08-24 16:27 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 13/30] linux-user/hexagon: Use force_sigsegv_code Richard Henderson
2021-08-24 16:29 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 14/30] linux-user/hppa: Use force_sig_fault, force_sigsegv_for_addr Richard Henderson
2021-08-24 16:32 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 15/30] linux-user/hppa: Use the proper si_code for PRIV_OPR, PRIV_REG, OVERFLOW Richard Henderson
2021-08-24 16:34 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 16/30] linux-user/hppa: Set FPE_CONDTRAP for COND Richard Henderson
2021-08-24 16:37 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 17/30] linux-user/i386: Split out maybe_handle_vm86_trap Richard Henderson
2021-08-24 16:38 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 18/30] linux-user/i386: Use force_sig, force_sig_fault, force_sigsegv_for_addr Richard Henderson
2021-08-24 16:40 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 19/30] linux-user/m68k: Use " Richard Henderson
2021-08-24 16:41 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 20/30] linux-user/microblaze: " Richard Henderson
2021-08-24 16:42 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 21/30] linux-user/microblaze: Fix SIGFPE si_codes Richard Henderson
2021-08-24 16:55 ` Peter Maydell [this message]
2021-08-22 3:55 ` [PATCH v2 22/30] linux-user/mips: Improve do_break Richard Henderson
2021-08-24 16:46 ` Philippe Mathieu-Daudé
2021-08-22 3:55 ` [PATCH v2 23/30] linux-user/mips: Use force_sig_fault, force_sigsegv_for_addr Richard Henderson
2021-08-24 17:04 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 24/30] linux-user/openrisc: " Richard Henderson
2021-08-24 17:17 ` Peter Maydell
2021-09-19 17:49 ` Richard Henderson
2021-09-21 20:26 ` Stafford Horne
2021-08-22 3:55 ` [PATCH v2 25/30] linux-user/ppc: " Richard Henderson
2021-08-24 17:19 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 26/30] linux-user/riscv: " Richard Henderson
2021-08-24 17:23 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 27/30] linux-user/s390x: " Richard Henderson
2021-08-24 17:23 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 28/30] linux-user/sh4: " Richard Henderson
2021-08-24 17:24 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 29/30] linux-user/sparc: " Richard Henderson
2021-08-24 17:25 ` Peter Maydell
2021-08-22 3:55 ` [PATCH v2 30/30] linux-user/xtensa: " Richard Henderson
2021-08-24 17:26 ` Peter Maydell
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='CAFEAcA9FiqbvCh1xZWdknEJV=jnoxgZqz=ky-A7zXMM+Re74AQ@mail.gmail.com' \
--to=peter.maydell@linaro.org \
--cc=edgar.iglesias@gmail.com \
--cc=laurent@vivier.eu \
--cc=qemu-devel@nongnu.org \
--cc=richard.henderson@linaro.org \
/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 NNTP newsgroup(s).