qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Richard Henderson <richard.henderson@linaro.org>
To: Kele Huang <kele.hwang@gmail.com>, qemu-devel@nongnu.org
Cc: Xu Zou <iwatchnima@gmail.com>
Subject: Re: [PATCH v2 1/1] accel/tcg: Fix computing of is_write for mips
Date: Thu, 24 Sep 2020 07:05:03 -0700	[thread overview]
Message-ID: <74c6f6d4-714a-4f8d-71fd-185cf3b35655@linaro.org> (raw)
In-Reply-To: <20200923093800.9845-1-kele.hwang@gmail.com>

On 9/23/20 2:38 AM, Kele Huang wrote:
> Detect mips store instructions in cpu_signal_handler for all MIPS
> versions, and set is_write if encountering such store instructions.
> 
> This fixed the error while dealing with self-modifed code for MIPS.
> 
> Signed-off-by: Kele Huang <kele.hwang@gmail.com>
> Signed-off-by: Xu Zou <iwatchnima@gmail.com>
> ---
>  accel/tcg/user-exec.c | 51 ++++++++++++++++++++++++++++++++++++++++++-
>  1 file changed, 50 insertions(+), 1 deletion(-)
> 
> diff --git a/accel/tcg/user-exec.c b/accel/tcg/user-exec.c
> index bb039eb32d..18784516e5 100644
> --- a/accel/tcg/user-exec.c
> +++ b/accel/tcg/user-exec.c
> @@ -710,11 +710,60 @@ int cpu_signal_handler(int host_signum, void *pinfo,
>      greg_t pc = uc->uc_mcontext.pc;
>      int is_write;
>  
> -    /* XXX: compute is_write */
>      is_write = 0;
> +
> +    /* Detect store by reading the instruction at the program counter. */
> +    uint32_t insn = *(uint32_t *)pc;
> +    switch(insn>>29) {

This would be easier if you simply looked at the entire major opcode field,
beginning at bit 26.

> +    case 0x5:
> +        switch((insn>>26) & 0x7) {
> +        case 0x0: /* SB */
> +        case 0x1: /* SH */
> +        case 0x2: /* SWL */
> +        case 0x3: /* SW */
> +        case 0x4: /* SDL */
> +        case 0x5: /* SDR */
> +        case 0x6: /* SWR */
> +            is_write = 1;
> +        }

So this becomes

    case 050: /* SB */
    case 051: /* SH */
      ...

I know there are some who don't like octal, but IMO MIPS and its 6 bit fields
and 8x8 tables is a natural fit -- one can read the two octal digits right off
of the table.

Otherwise, perhaps you'd prefer binary constants like 0b101000.  But with these
tables I find the mental bit-extract from hex to be tiresome.

> +        break;
> +    case 0x7:
> +        switch((insn>>26) & 0x7) {
> +        case 0x0: /* SC */
> +        case 0x1: /* SWC1 */
> +        case 0x4: /* SCD */
> +        case 0x5: /* SDC1 */
> +        case 0x7: /* SD */
> +#if !defined(__mips_isa_rev) || __mips_isa_rev < 6
> +        case 0x2: /* SWC2 */
> +        case 0x6: /* SDC2 */
> +#endif
> +            is_write = 1;

Similarly.

> +        }
> +        break;
> +    }
> +
> +    /*
> +     * Required in all versions of MIPS64 since MIPS64r1. Not available
> +     * in MIPS32r1. Required by MIPS32r2 and subsequent versions of MIPS32.
> +     */
> +    switch ((insn >> 3) & 0x7) {
> +    case 0x1:
> +        switch (insn & 0x7) {
> +        case 0x0: /* SWXC1 */
> +        case 0x1: /* SDXC1 */
> +            is_write = 1;
> +        }
> +        break;
> +    }

This switch is incorrectly placed.  It must be within the first switch, under
major opcode 023 (COP1X).  And again, you should extract the entire 6-bit minor
opcode all at once, not one octal digit at a time.

> +#elif defined(__misp16) || defined(__mips_micromips)
> +
> +#error "Unsupported encoding"

This is incorrectly placed, because we've already successfully entered the
preceeding #elif defined(__mips__).  This needs to be

#elif defined(__mips__)
# if defined(__mips16) || defined(__mips_micromips)
#  error
# endif

int cpu_signal_handler(int host_signum, void *pinfo,
                       void *puc)
{
   ...
}

#elif defined(__riscv)



r~


  parent reply	other threads:[~2020-09-24 14:06 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <503406>
2020-09-10  7:43 ` [PATCH 0/1] accel/tcg: Fix computing is_write for mips Kele Huang
2020-09-10  7:43   ` [PATCH 1/1] " Kele Huang
2020-09-10 17:18     ` Richard Henderson
2020-09-24  8:59       ` Kele Huang
2020-09-23  9:38 ` [PATCH v2 1/1] accel/tcg: Fix computing of " Kele Huang
2020-09-23 11:08   ` Philippe Mathieu-Daudé
2020-09-24  8:52     ` Kele Huang
2020-09-24 10:01     ` Kele Huang
2020-09-24 14:05   ` Richard Henderson [this message]
2020-09-25  8:34     ` Kele Huang

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=74c6f6d4-714a-4f8d-71fd-185cf3b35655@linaro.org \
    --to=richard.henderson@linaro.org \
    --cc=iwatchnima@gmail.com \
    --cc=kele.hwang@gmail.com \
    --cc=qemu-devel@nongnu.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).