From: Andrew Cooper <andrew.cooper3@citrix.com>
To: Jan Beulich <JBeulich@suse.com>
Cc: xen-devel <xen-devel@lists.xenproject.org>,
Keir Fraser <keir@xen.org>, Eddie Dong <eddie.dong@intel.com>,
Jun Nakajima <jun.nakajima@intel.com>
Subject: Re: [PATCH v2 2/4] VMX: move various uses of UD2 out of fast paths
Date: Mon, 26 Aug 2013 10:09:32 +0100 [thread overview]
Message-ID: <521B1B4C.3040505@citrix.com> (raw)
In-Reply-To: <521B34C402000078000EE476@nat28.tlf.novell.com>
[-- Attachment #1.1: Type: text/plain, Size: 4427 bytes --]
On 26/08/2013 09:58, Jan Beulich wrote:
> ... at once making conditional forward jumps, which are statically
> predicted to be not taken, only used for the unlikely (error) cases.
>
> Signed-off-by: Jan Beulich <jbeulich@suse.com>
Reviewed-by: Andrew Cooper <andrew.cooper3@citrix.com>
> ---
> v2: Fix #UD recovery for INVVPID (unintended flow change spotted by Andrew Cooper).
>
> --- a/xen/include/asm-x86/asm_defns.h
> +++ b/xen/include/asm-x86/asm_defns.h
> @@ -67,6 +67,30 @@ void ret_from_intr(void);
> #define ASSERT_NOT_IN_ATOMIC
> #endif
>
> +#else
> +
> +#ifdef __clang__ /* clang's builtin assember can't do .subsection */
> +
> +#define UNLIKELY_START_SECTION ".pushsection .fixup,\"ax\""
> +#define UNLIKELY_END_SECTION ".popsection"
> +
> +#else
> +
> +#define UNLIKELY_START_SECTION ".subsection 1"
> +#define UNLIKELY_END_SECTION ".subsection 0"
> +
> +#endif
> +
> +#define UNLIKELY_START(cond, tag) \
> + "j" #cond " .Lunlikely%=.tag;\n\t" \
> + UNLIKELY_START_SECTION "\n" \
> + ".Lunlikely%=.tag:"
> +
> +#define UNLIKELY_END(tag) \
> + "jmp .Llikely%=.tag;\n\t" \
> + UNLIKELY_END_SECTION "\n" \
> + ".Llikely%=.tag:"
> +
> #endif
>
> #endif /* __X86_ASM_DEFNS_H__ */
> --- a/xen/include/asm-x86/hvm/vmx/vmx.h
> +++ b/xen/include/asm-x86/hvm/vmx/vmx.h
> @@ -285,7 +285,9 @@ static inline void __vmptrld(u64 addr)
> asm volatile ( VMPTRLD_OPCODE
> MODRM_EAX_06
> /* CF==1 or ZF==1 --> crash (ud2) */
> - "ja 1f ; ud2 ; 1:\n"
> + UNLIKELY_START(be, vmptrld)
> + "\tud2\n"
> + UNLIKELY_END_SECTION
> :
> : "a" (&addr)
> : "memory");
> @@ -296,7 +298,9 @@ static inline void __vmpclear(u64 addr)
> asm volatile ( VMCLEAR_OPCODE
> MODRM_EAX_06
> /* CF==1 or ZF==1 --> crash (ud2) */
> - "ja 1f ; ud2 ; 1:\n"
> + UNLIKELY_START(be, vmclear)
> + "\tud2\n"
> + UNLIKELY_END_SECTION
> :
> : "a" (&addr)
> : "memory");
> @@ -309,7 +313,9 @@ static inline unsigned long __vmread(uns
> asm volatile ( VMREAD_OPCODE
> MODRM_EAX_ECX
> /* CF==1 or ZF==1 --> crash (ud2) */
> - "ja 1f ; ud2 ; 1:\n"
> + UNLIKELY_START(be, vmread)
> + "\tud2\n"
> + UNLIKELY_END_SECTION
> : "=c" (ecx)
> : "a" (field)
> : "memory");
> @@ -322,7 +328,9 @@ static inline void __vmwrite(unsigned lo
> asm volatile ( VMWRITE_OPCODE
> MODRM_EAX_ECX
> /* CF==1 or ZF==1 --> crash (ud2) */
> - "ja 1f ; ud2 ; 1:\n"
> + UNLIKELY_START(be, vmwrite)
> + "\tud2\n"
> + UNLIKELY_END_SECTION
> :
> : "a" (field) , "c" (value)
> : "memory");
> @@ -360,7 +368,9 @@ static inline void __invept(int type, u6
> asm volatile ( INVEPT_OPCODE
> MODRM_EAX_08
> /* CF==1 or ZF==1 --> crash (ud2) */
> - "ja 1f ; ud2 ; 1:\n"
> + UNLIKELY_START(be, invept)
> + "\tud2\n"
> + UNLIKELY_END_SECTION
> :
> : "a" (&operand), "c" (type)
> : "memory" );
> @@ -377,7 +387,10 @@ static inline void __invvpid(int type, u
> /* Fix up #UD exceptions which occur when TLBs are flushed before VMXON. */
> asm volatile ( "1: " INVVPID_OPCODE MODRM_EAX_08
> /* CF==1 or ZF==1 --> crash (ud2) */
> - "ja 2f ; ud2 ; 2:\n"
> + UNLIKELY_START(be, invvpid)
> + "\tud2\n"
> + UNLIKELY_END_SECTION "\n"
> + "2:"
> _ASM_EXTABLE(1b, 2b)
> :
> : "a" (&operand), "c" (type)
>
>
>
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel
[-- Attachment #1.2: Type: text/html, Size: 5167 bytes --]
[-- Attachment #2: Type: text/plain, Size: 126 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel
next prev parent reply other threads:[~2013-08-26 9:09 UTC|newest]
Thread overview: 60+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-08-23 13:58 [PATCH 0/4] HVM: produce better binary code Jan Beulich
2013-08-23 14:01 ` [PATCH 1/4] VMX: streamline entry.S code Jan Beulich
2013-08-26 10:44 ` Andrew Cooper
2013-08-26 11:01 ` Jan Beulich
2013-08-26 11:48 ` Andrew Cooper
2013-08-26 13:12 ` Jan Beulich
2013-08-26 13:22 ` Andrew Cooper
2013-08-29 11:01 ` Tim Deegan
2013-08-29 12:35 ` Jan Beulich
2013-08-23 14:02 ` [PATCH 2/4] VMX: move various uses of UD2 out of fast paths Jan Beulich
2013-08-23 22:06 ` Andrew Cooper
2013-08-26 8:50 ` Jan Beulich
2013-08-26 9:07 ` Andrew Cooper
2013-08-26 8:58 ` [PATCH v2 " Jan Beulich
2013-08-26 9:09 ` Andrew Cooper [this message]
2013-08-29 11:08 ` Tim Deegan
2013-08-23 14:03 ` [PATCH 3/4] VMX: use proper instruction mnemonics if assembler supports them Jan Beulich
2013-08-24 22:18 ` Andrew Cooper
2013-08-26 9:06 ` Jan Beulich
2013-08-26 9:25 ` Andrew Cooper
2013-08-26 9:41 ` Jan Beulich
2013-08-26 10:18 ` [PATCH v3 " Jan Beulich
2013-08-26 13:05 ` Andrew Cooper
2013-08-26 13:20 ` Jan Beulich
2013-08-26 14:03 ` [PATCH v4 " Jan Beulich
2013-08-26 14:18 ` Andrew Cooper
2013-08-26 14:29 ` Jan Beulich
2013-08-26 15:07 ` Andrew Cooper
2013-08-26 15:10 ` Andrew Cooper
2013-08-26 15:30 ` Jan Beulich
2013-08-26 15:29 ` Jan Beulich
2013-08-26 15:33 ` Andrew Cooper
2013-08-26 15:31 ` [PATCH v5 " Jan Beulich
2013-08-26 15:36 ` Andrew Cooper
2013-08-29 11:47 ` Tim Deegan
2013-08-29 12:30 ` Jan Beulich
2013-08-29 13:11 ` Tim Deegan
2013-08-29 13:27 ` Jan Beulich
2013-08-29 14:02 ` Tim Deegan
2013-08-29 12:45 ` Jan Beulich
2013-08-29 13:19 ` Tim Deegan
2013-08-26 9:03 ` [PATCH v2 " Jan Beulich
2013-08-23 14:04 ` [PATCH 4/4] SVM: streamline entry.S code Jan Beulich
2013-08-26 16:20 ` Andrew Cooper
2013-08-26 17:20 ` Keir Fraser
2013-08-26 17:46 ` Andrew Cooper
2013-08-26 21:47 ` Andrew Cooper
2013-08-27 7:38 ` Jan Beulich
2013-08-29 11:56 ` Tim Deegan
2013-09-04 14:39 ` Boris Ostrovsky
2013-09-04 14:50 ` Jan Beulich
2013-09-04 15:09 ` Boris Ostrovsky
2013-09-04 15:20 ` Jan Beulich
2013-09-04 16:42 ` Boris Ostrovsky
2013-09-05 7:10 ` Jan Beulich
2013-09-04 10:06 ` Ping: [PATCH 0/4] HVM: produce better binary code Jan Beulich
2013-09-04 16:16 ` Andrew Cooper
2013-09-04 16:30 ` Tim Deegan
2013-09-05 7:52 ` Jan Beulich
2013-09-05 7:58 ` Tim Deegan
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=521B1B4C.3040505@citrix.com \
--to=andrew.cooper3@citrix.com \
--cc=JBeulich@suse.com \
--cc=eddie.dong@intel.com \
--cc=jun.nakajima@intel.com \
--cc=keir@xen.org \
--cc=xen-devel@lists.xenproject.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 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.