kvm.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jim Mattson <jmattson@google.com>
To: Bill Wendling <morbo@google.com>
Cc: kvm list <kvm@vger.kernel.org>,
	Paolo Bonzini <pbonzini@redhat.com>,
	alexandru.elisei@arm.com
Subject: Re: [kvm-unit-tests PATCH 1/2] x86: realmode: explicitly copy structure to avoid memcpy
Date: Wed, 16 Oct 2019 12:07:50 -0700	[thread overview]
Message-ID: <CALMp9eSAG_1cFGG65hpbOpcSLi8MJRwmdXnkeaMAqfLee6faQQ@mail.gmail.com> (raw)
In-Reply-To: <20191012235859.238387-2-morbo@google.com>

On Sat, Oct 12, 2019 at 4:59 PM Bill Wendling <morbo@google.com> wrote:
>
> Clang prefers to use a "mempcy" (or equivalent) to copy the "regs"
> structure. This doesn't work in 16-bit mode, as it will end up copying
> over half the number of bytes. GCC performs a field-by-field copy of the
> structure, so force clang to do the same thing.
>
> Signed-off-by: Bill Wendling <morbo@google.com>
> ---
>  x86/realmode.c | 19 ++++++++++++++++---
>  1 file changed, 16 insertions(+), 3 deletions(-)
>
> diff --git a/x86/realmode.c b/x86/realmode.c
> index 303d093..cf45fd6 100644
> --- a/x86/realmode.c
> +++ b/x86/realmode.c
> @@ -117,6 +117,19 @@ struct regs {
>         u32 eip, eflags;
>  };
>
> +#define COPY_REG(name, dst, src) (dst).name = (src).name
> +#define COPY_REGS(dst, src)                            \
> +       COPY_REG(eax, dst, src);                        \
> +       COPY_REG(ebx, dst, src);                        \
> +       COPY_REG(ecx, dst, src);                        \
> +       COPY_REG(edx, dst, src);                        \
> +       COPY_REG(esi, dst, src);                        \
> +       COPY_REG(edi, dst, src);                        \
> +       COPY_REG(esp, dst, src);                        \
> +       COPY_REG(ebp, dst, src);                        \
> +       COPY_REG(eip, dst, src);                        \
> +       COPY_REG(eflags, dst, src)
> +

This seems very fragile, too. Can we introduce our own
address-space-size-independent "memcpy" and use that?

I'm thinking something like:

static void bytecopy(void *dst, void *src, u32 count)
{
        asm volatile("rep movsb"
             : "+D" (dst), "+S" (src), "+c" (count) : : "cc");
}

>  struct table_descr {
>         u16 limit;
>         void *base;
> @@ -148,11 +161,11 @@ static void exec_in_big_real_mode(struct insn_desc *insn)
>         extern u8 test_insn[], test_insn_end[];
>
>         for (i = 0; i < insn->len; ++i)
> -           test_insn[i] = ((u8 *)(unsigned long)insn->ptr)[i];
> +               test_insn[i] = ((u8 *)(unsigned long)insn->ptr)[i];
>         for (; i < test_insn_end - test_insn; ++i)
>                 test_insn[i] = 0x90; // nop
>
> -       save = inregs;
> +       COPY_REGS(save, inregs);
>         asm volatile(
>                 "lgdtl %[gdt_descr] \n\t"
>                 "mov %%cr0, %[tmp] \n\t"
> @@ -196,7 +209,7 @@ static void exec_in_big_real_mode(struct insn_desc *insn)
>                 : [gdt_descr]"m"(gdt_descr), [bigseg]"r"((short)16)
>                 : "cc", "memory"
>                 );
> -       outregs = save;
> +       COPY_REGS(outregs, save);
>  }
>
>  #define R_AX 1
> --
> 2.23.0.700.g56cf767bdb-goog
>

  reply	other threads:[~2019-10-16 19:08 UTC|newest]

Thread overview: 14+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-10-12 23:58 [kvm-unit-tests PATCH 0/2] realmode test fixes for clang Bill Wendling
2019-10-12 23:58 ` [kvm-unit-tests PATCH 1/2] x86: realmode: explicitly copy structure to avoid memcpy Bill Wendling
2019-10-16 19:07   ` Jim Mattson [this message]
2019-10-21 15:38   ` Paolo Bonzini
2019-10-12 23:58 ` [kvm-unit-tests PATCH 2/2] x86: realmode: use inline asm to get stack pointer Bill Wendling
2019-10-16 19:53   ` Jim Mattson
2019-10-16 21:52     ` Jim Mattson
2019-10-21 15:41       ` Paolo Bonzini
2019-10-17  1:25 ` [kvm-unit-tests v2 PATCH 0/2] realmode test fixes for clang Bill Wendling
2019-10-17  1:25   ` [kvm-unit-tests v2 PATCH 1/2] x86: realmode: explicitly copy regs structure Bill Wendling
2019-10-17  1:25   ` [kvm-unit-tests v2 PATCH 2/2] x86: realmode: fix esp in call test Bill Wendling
2019-10-17 23:27     ` Jim Mattson
2019-10-21 15:43     ` Paolo Bonzini
2019-10-21 16:43       ` Jim Mattson

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=CALMp9eSAG_1cFGG65hpbOpcSLi8MJRwmdXnkeaMAqfLee6faQQ@mail.gmail.com \
    --to=jmattson@google.com \
    --cc=alexandru.elisei@arm.com \
    --cc=kvm@vger.kernel.org \
    --cc=morbo@google.com \
    --cc=pbonzini@redhat.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 NNTP newsgroup(s).