All of lore.kernel.org
 help / color / mirror / Atom feed
From: Sean Christopherson <seanjc@google.com>
To: Uros Bizjak <ubizjak@gmail.com>
Cc: kvm@vger.kernel.org, linux-kernel@vger.kernel.org,
	Paolo Bonzini <pbonzini@redhat.com>
Subject: Re: [PATCH] KVM: x86: Improve exception safe wrappers in emulate.c
Date: Wed, 13 Oct 2021 19:24:40 +0000	[thread overview]
Message-ID: <YWcyeGk7vOSoQWW4@google.com> (raw)
In-Reply-To: <20210917135152.5111-1-ubizjak@gmail.com>

On Fri, Sep 17, 2021, Uros Bizjak wrote:
> Improve exception safe wrappers in emulate.c by converting them to
> ASM GOTO (and ASM GOTO OUTPUT when supported) statements.  Also, convert
> wrappers to inline functions to avoid statement as expression
> GNU extension and to remove weird requirement where user must know
> where the asm argument is being expanded.
> 
> Cc: Paolo Bonzini <pbonzini@redhat.com>
> Cc: Sean Christopherson  <seanjc@google.com>
> Signed-off-by: Uros Bizjak <ubizjak@gmail.com>
> ---
>  arch/x86/kvm/emulate.c | 80 ++++++++++++++++++++++++++++++------------
>  1 file changed, 57 insertions(+), 23 deletions(-)
> 
> diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
> index 2837110e66ed..2197a3ecc55b 100644
> --- a/arch/x86/kvm/emulate.c
> +++ b/arch/x86/kvm/emulate.c
> @@ -464,25 +464,59 @@ FOP_FUNC(salc)
>  FOP_RET(salc)
>  FOP_END;
>  
> -/*
> - * XXX: inoutclob user must know where the argument is being expanded.

I 100% agree that this is a weird requirement, but I actually like the side
effect of forcing the caller to define a name for the input/output.

> - *      Relying on CONFIG_CC_HAS_ASM_GOTO would allow us to remove _fault.
> - */
> -#define asm_safe(insn, inoutclob...) \
> -({ \
> -	int _fault = 0; \
> - \
> -	asm volatile("1:" insn "\n" \
> -	             "2:\n" \
> -	             ".pushsection .fixup, \"ax\"\n" \
> -	             "3: movl $1, %[_fault]\n" \
> -	             "   jmp  2b\n" \
> -	             ".popsection\n" \
> -	             _ASM_EXTABLE(1b, 3b) \
> -	             : [_fault] "+qm"(_fault) inoutclob ); \
> - \
> -	_fault ? X86EMUL_UNHANDLEABLE : X86EMUL_CONTINUE; \
> -})
> +static __always_inline int safe_fwait(void)
> +{
> +	asm_volatile_goto("1: fwait\n\t"
> +			  _ASM_EXTABLE(1b, %l[fault])
> +			  : : : : fault);
> +	return X86EMUL_CONTINUE;
> + fault:
> +	return X86EMUL_UNHANDLEABLE;
> +}

Rather than defining a bunch of safe_() variants, what about providing a generic
helper/macro similar to the existing asm_safe()?  Not just for KVM, but for the
kernel at large.  Asm with output is problematic due to the CONFIG_CC_HAS_ASM_GOTO
dependency, but it wouldn't be the end of the world to state that simply isn't
supported until the min compiler version is raised.

__wrmsr(), native_write_msr_safe(), cpu_vmxoff(), kvm_cpu_vmxon(), and probably
others could use a generic generic asm_safe().  I wouldn't be surprised if there
are other places in the kernel that could take advantage of such a helper, e.g.
kvm_load_ldt() could use a "safe" variant instead of crashing if the sel is bad.

  reply	other threads:[~2021-10-13 19:24 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-09-17 13:51 [PATCH] KVM: x86: Improve exception safe wrappers in emulate.c Uros Bizjak
2021-10-13 19:24 ` Sean Christopherson [this message]
2021-11-01  9:59   ` Uros Bizjak

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=YWcyeGk7vOSoQWW4@google.com \
    --to=seanjc@google.com \
    --cc=kvm@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=pbonzini@redhat.com \
    --cc=ubizjak@gmail.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.