qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Laurent Vivier <laurent@vivier.eu>
To: Max Filippov <jcmvbkbc@gmail.com>,
	Richard Henderson <richard.henderson@linaro.org>
Cc: qemu-devel@nongnu.org, Peter Maydell <peter.maydell@linaro.org>,
	Riku Voipio <riku.voipio@iki.fi>
Subject: Re: [Qemu-devel] [PATCH v3] linux-user: fix preadv/pwritev offsets
Date: Thu, 5 Apr 2018 17:52:39 +0200	[thread overview]
Message-ID: <6b21a387-7ebd-c738-e1c4-b65645f11fed@vivier.eu> (raw)
In-Reply-To: <20180405134702.5020-1-jcmvbkbc@gmail.com>

Le 05/04/2018 à 15:47, Max Filippov a écrit :
> preadv/pwritev accept low and high parts of file offset in two separate
> parameters. When host bitness doesn't match guest bitness these parts
> must be appropriately recombined.
> Introduce target_to_host_low_high that does this recombination and use
> it in preadv/pwritev syscalls.
> 
> This fixes glibc testsuite test misc/tst-preadvwritev64.
> 
> Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
> ---
> Changes v2->v3:
> - rename helper to target_to_host_low_high;
> - handle cases TARGET_LONG_BITS * 2 > HOST_LONG_BITS and
>   TARGET_LONG_BITS < 2 * HOST_LONG_BITS correctly.
> 
> Changes v1->v2:
> - fix host high computation in TARGET_LONG_BITS > HOST_LONG_BITS case
> 
>  linux-user/syscall.c | 34 ++++++++++++++++++++++++++++++++--
>  1 file changed, 32 insertions(+), 2 deletions(-)
> 
> diff --git a/linux-user/syscall.c b/linux-user/syscall.c
> index 5ef517613577..13ad572e0d3b 100644
> --- a/linux-user/syscall.c
> +++ b/linux-user/syscall.c
> @@ -3386,6 +3386,30 @@ static abi_long do_getsockopt(int sockfd, int level, int optname,
>      return ret;
>  }
>  
> +static void target_to_host_low_high(abi_ulong tlow,
> +                                    abi_ulong thigh,
> +                                    unsigned long *hlow,
> +                                    unsigned long *hhigh)
> +{
> +#if TARGET_LONG_BITS == HOST_LONG_BITS
> +        *hlow = tlow;
> +        *hhigh = thigh;
> +#elif TARGET_LONG_BITS < HOST_LONG_BITS
> +        *hlow = tlow | (unsigned long)thigh << TARGET_LONG_BITS;
> +#if TARGET_LONG_BITS * 2 <= HOST_LONG_BITS
> +        *hhigh = 0;
> +#else
> +        *hhigh = (unsigned long)thigh >> (HOST_LONG_BITS - TARGET_LONG_BITS);
> +#endif
> +#else
> +        *hlow = (unsigned long)tlow;
> +        *hhigh = (unsigned long)(tlow >> HOST_LONG_BITS);
> +#if TARGET_LONG_BITS < 2 * HOST_LONG_BITS
> +        *hhigh |= (unsigned long)thigh << (TARGET_LONG_BITS - HOST_LONG_BITS);
> +#endif
> +#endif
> +}

Why don't you try to de-construct then re-construct the offset?

Kernel commit
  601cc11d054a "Make non-compat preadv/pwritev use native register size"
is interesting.

static inline abi_llong target_pos_from_hilo(abi_ulong high,
                                             abi_ulong low)
{
#define TARGET_HALF_LONG_BITS (TARGET_LONG_BITS / 2)
   return (((abi_llong)high << TARGET_HALF_LONG_BITS) <<
                             TARGET_HALF_LONG_BITS) | low;
}

#define HOST_HALF_LONG_BITS (HOST_LONG_BITS / 2)

 abi_llong pos = target_pos_from_hilo(arg4, arg5);
 ret = get_errno(safe_preadv(arg1, vec, arg3,
          pos,
          (pos >> HOST_HALF_LONG_BITS) >> HOST_HALF_LONG_BITS));

It looks simpler, but perhaps I miss something
(it's just cut&paste, I don't pretend to understand that code...)?

Richard?

Thanks,
Laurent

  reply	other threads:[~2018-04-05 15:52 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-04-05 13:47 [Qemu-devel] [PATCH v3] linux-user: fix preadv/pwritev offsets Max Filippov
2018-04-05 15:52 ` Laurent Vivier [this message]
2018-04-05 16:27   ` Max Filippov
2018-04-05 16:33     ` Laurent Vivier

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=6b21a387-7ebd-c738-e1c4-b65645f11fed@vivier.eu \
    --to=laurent@vivier.eu \
    --cc=jcmvbkbc@gmail.com \
    --cc=peter.maydell@linaro.org \
    --cc=qemu-devel@nongnu.org \
    --cc=richard.henderson@linaro.org \
    --cc=riku.voipio@iki.fi \
    /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).