From: "H.J. Lu" <hjl.tools@gmail.com>
To: Mark Kettenis <mark.kettenis@xs4all.nl>
Cc: gdb-patches@sourceware.org
Subject: Re: PATCH: 0/6 [2nd try]: Add AVX support
Date: Sun, 07 Mar 2010 16:40:00 -0000 [thread overview]
Message-ID: <6dc9ffc81003070840x649dcc23qa088fddd69df7e54@mail.gmail.com> (raw)
In-Reply-To: <6dc9ffc81003070831yf923b9bj83da56cb9c83a751@mail.gmail.com>
On Sun, Mar 7, 2010 at 8:31 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
> On Sun, Mar 7, 2010 at 6:37 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
>> On Sun, Mar 7, 2010 at 6:16 AM, Mark Kettenis <mark.kettenis@xs4all.nl> wrote:
>>>> Date: Sat, 6 Mar 2010 14:16:34 -0800
>>>> From: "H.J. Lu" <hongjiu.lu@intel.com>
>>>>
>>>> AVX registers are saved and restored via the XSAVE extended state. The
>>>> extended control register 0 (the XFEATURE_ENABLED_MASK register), XCR0,
>>>> is used to determine which states, x87, SSE, AVX, ... are supported
>>>> in the XSAVE extended state. XCR0 can be read with the new "xgetbv"
>>>> instruction. The xstate_bv field at byte offset 512 in the XSAVE
>>>> extended state indicates what states the current process is in. If
>>>> the feature bit is cleared, the corresponding registers should be read as
>>>> 0. If we update a register, we should set the corresponding feature
>>>> bit in the xstate_bv field.
>>>>
>>>> We added PTRACE_GETREGSET and PTRACE_SETREGSET to Linux kernel to
>>>> fetch and store AVX registers with ptrace. Linux kernel also stores
>>>> XCR0 at the first 8 bytes of the software usable bytes, starting at
>>>> byte offset 464.
>>>>
>>>> There are total 6 patches to add AVX support for Linux. They support:
>>>>
>>>> 1. The upper 128bit YMM registers are added for AVX support. The upper
>>>> 128bit YMM registers are hidden from users. Gdb combines XMM register,
>>>> %xmmX, with 128bit YMM register, %ymmXh, and present the whole 256bit
>>>> YMM register, %ymmX, as pseudo register to users.
>>>> 2. Backward compatible. If AVX isn't supported, SSE will be used.
>>>> 3. Forward compatible. If new state beyond AVX is supported in
>>>> the XSAVE extended state, only AVX state will be used.
>>>> 4. Remote gdb protocol extension. GDB will send "x86=xml" in qSupported
>>>> request packet to indicate that GDB supports x86 XML target desciption.
>>>> The gdb stub will send x86 XML target desciption if it sees "x86=xml"
>>>> in qSupported request packet.
>>>>
>>>> One advantage of this approach is YMM registers are actually stored as
>>>> XMM registers and upper YMM registers in the XSAVE extended state. It
>>>> is easy and natural to access them as %xmmX and %ymmXh internally. We
>>>> just need to hide %ymmXh from users.
>>>>
>>>> To support AVX on other OSes, the following changes are needed:
>>>>
>>>> 1. Kernel support to get/set the XSAVE extended state.
>>>> 2. Handle 8/16 upper YMM registers.
>>>> 3. Provide target to_read_description to return SSE or AVX target
>>>> description.
>>>> 4. Update gdbarch_core_read_description to return SSE or AVX target
>>>> description based on contents of core dump.
>>>
>>> Wait; there is something important missing here. How are the new %ymm
>>> registers referred to in debug info? The AMD64 SysV psABI defines the
>>> DWARF register Number Mapping, but the 0.99.4 draft copy I have
>>> doesn't define any mappings for the %ymm registers. What mapping does
>>> GCC use?
>>>
>>
>> In gcc, XMM and YMM registers have the same register number. They map
>> to be the same DWARF register with different sizes. Since XMM and YMM
>> registers are caller-saved, they don't appear in unwind info. So, the same
>> DWARF register with different sizes for XMM/YMM registers isn't a problem.
>>
>>
>
> Yes, there is a problem. amd64_dwarf_reg_to_regnum needs to map 256bit
> register to YMM. How do other arches solve this?
>
My first approach works here since XMM and YMM register have the same
register numbers. We can solve it with 2 alternatives:
1. Give a different DWARF register number for YMM register,
which is an incompatible ABI change.
2. Implement YMM registers as a super set of XMM registers, which
is my first approach.
Thanks Mark for pointing out this issue.
--
H.J.
next prev parent reply other threads:[~2010-03-07 16:40 UTC|newest]
Thread overview: 115+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-03-04 18:02 PATCH: 1/6: " H.J. Lu
2010-03-04 18:05 ` PATCH: 2/6: Add AVX support (Update document) H.J. Lu
2010-03-04 18:06 ` PATCH: 3/6: Add AVX support (i386 changes) H.J. Lu
2010-03-06 22:21 ` PATCH: 3/6 [2nd try]: " H.J. Lu
2010-03-07 21:32 ` H.J. Lu
2010-03-11 22:37 ` Mark Kettenis
2010-03-12 0:00 ` H.J. Lu
2010-03-27 14:55 ` Mark Kettenis
2010-03-27 15:30 ` Daniel Jacobowitz
2010-03-27 16:05 ` Mark Kettenis
2010-03-27 15:33 ` H.J. Lu
2010-03-27 16:09 ` Mark Kettenis
2010-03-28 1:39 ` H.J. Lu
2010-03-12 16:49 ` H.J. Lu
2010-03-13 1:38 ` H.J. Lu
2010-03-29 1:11 ` PATCH: 3/6 [3rd " H.J. Lu
2010-04-02 14:31 ` H.J. Lu
2010-04-02 14:42 ` Mark Kettenis
2010-04-02 15:28 ` H.J. Lu
2010-04-07 10:13 ` Mark Kettenis
2010-04-07 14:56 ` H.J. Lu
2010-04-07 15:04 ` H.J. Lu
2010-04-07 15:19 ` Mark Kettenis
2010-04-07 16:55 ` H.J. Lu
2010-04-07 18:34 ` Mark Kettenis
2010-04-07 18:50 ` H.J. Lu
2010-03-27 15:48 ` PATCH: 3/6 [2nd " Mark Kettenis
2010-03-28 1:37 ` H.J. Lu
2010-03-28 11:55 ` Mark Kettenis
2010-03-28 14:25 ` H.J. Lu
2010-03-29 20:32 ` Mark Kettenis
2010-03-29 21:41 ` H.J. Lu
2010-03-04 18:08 ` PATCH: 4/6: Add AVX support (amd64 changes) H.J. Lu
2010-03-04 18:09 ` PATCH: 5/6: Add AVX support (i387 changes) H.J. Lu
2010-03-04 18:10 ` PATCH: 6/6: Add AVX support (gdbserver changes) H.J. Lu
2010-03-06 22:23 ` PATCH: 6/6 [2nd try]: " H.J. Lu
2010-03-12 17:25 ` H.J. Lu
2010-03-27 16:07 ` Daniel Jacobowitz
2010-03-28 1:11 ` H.J. Lu
2010-03-28 7:55 ` Pedro Alves
2010-03-28 14:56 ` H.J. Lu
2010-03-28 16:17 ` Pedro Alves
2010-03-28 16:37 ` H.J. Lu
2010-03-28 16:40 ` Daniel Jacobowitz
2010-03-28 16:47 ` Pedro Alves
2010-03-28 20:53 ` H.J. Lu
2010-03-28 21:27 ` Pedro Alves
2010-03-28 16:39 ` Daniel Jacobowitz
2010-03-28 19:31 ` H.J. Lu
2010-03-29 1:09 ` PATCH: 6/6 [3rd " H.J. Lu
2010-03-29 14:08 ` Eli Zaretskii
2010-03-29 14:42 ` H.J. Lu
2010-03-29 15:11 ` Eli Zaretskii
2010-03-29 15:42 ` H.J. Lu
2010-03-29 15:51 ` Eli Zaretskii
2010-03-30 16:48 ` H.J. Lu
2010-04-02 17:39 ` Daniel Jacobowitz
2010-04-07 4:37 ` H.J. Lu
2010-04-03 21:57 ` Jan Kratochvil
2010-04-07 4:12 ` H.J. Lu
2010-04-07 16:59 ` H.J. Lu
2010-03-05 3:20 ` PATCH: 5/6: Add AVX support (i387 changes) Hui Zhu
2010-03-05 3:54 ` H.J. Lu
2010-03-06 22:22 ` PATCH: 5/6 [2nd try]: " H.J. Lu
2010-03-12 17:24 ` H.J. Lu
2010-04-07 16:57 ` PATCH: 5/6 [3rd " H.J. Lu
2010-03-27 15:08 ` PATCH: 5/6 [2nd " Mark Kettenis
2010-03-27 15:15 ` H.J. Lu
2010-03-06 22:21 ` PATCH: 4/6 [2nd try]: Add AVX support (amd64 changes) H.J. Lu
2010-03-07 21:33 ` H.J. Lu
2010-03-12 17:01 ` H.J. Lu
2010-03-13 1:38 ` H.J. Lu
2010-03-29 1:07 ` PATCH: 4/6 [3rd " H.J. Lu
2010-04-02 14:32 ` H.J. Lu
2010-04-07 16:54 ` H.J. Lu
2010-03-05 10:33 ` PATCH: 2/6: Add AVX support (Update document) Eli Zaretskii
2010-03-05 14:08 ` H.J. Lu
2010-03-06 22:19 ` PATCH: 2/6 [2nd try]: " H.J. Lu
2010-03-12 11:11 ` Eli Zaretskii
2010-03-12 14:17 ` H.J. Lu
2010-03-12 15:28 ` Eli Zaretskii
2010-03-12 15:27 ` Eli Zaretskii
2010-03-12 16:46 ` H.J. Lu
2010-03-12 18:15 ` Eli Zaretskii
2010-03-29 0:18 ` PATCH: 2/6 [3rd " H.J. Lu
2010-03-30 16:41 ` H.J. Lu
2010-03-30 18:27 ` Eli Zaretskii
2010-03-30 18:37 ` H.J. Lu
2010-03-04 19:09 ` PATCH: 1/6: Add AVX support Daniel Jacobowitz
2010-03-04 19:29 ` H.J. Lu
2010-03-04 19:47 ` Daniel Jacobowitz
2010-03-04 21:27 ` H.J. Lu
2010-03-04 21:34 ` Nathan Froyd
2010-03-04 21:41 ` H.J. Lu
2010-03-04 21:59 ` Nathan Froyd
2010-03-04 21:47 ` Daniel Jacobowitz
2010-03-05 2:06 ` H.J. Lu
2010-03-05 7:29 ` Mark Kettenis
2010-03-06 22:16 ` PATCH: 0/6 [2nd try]: " H.J. Lu
2010-03-06 22:18 ` PATCH: 1/6 [2nd try]: Add AVX support (AVX XML files) H.J. Lu
2010-03-07 14:16 ` PATCH: 0/6 [2nd try]: Add AVX support Mark Kettenis
2010-03-07 14:37 ` H.J. Lu
2010-03-07 16:31 ` H.J. Lu
2010-03-07 16:40 ` H.J. Lu [this message]
2010-03-07 17:04 ` H.J. Lu
2010-03-07 17:39 ` H.J. Lu
2010-03-07 20:00 ` Mark Kettenis
2010-03-07 19:10 ` Nathan Froyd
2010-03-07 19:49 ` Mark Kettenis
2010-03-07 21:07 ` Nathan Froyd
2010-03-07 21:17 ` H.J. Lu
2010-03-07 20:29 ` Mark Kettenis
2010-03-07 21:04 ` H.J. Lu
2010-03-27 16:16 ` Daniel Jacobowitz
2010-03-29 0:16 ` PATCH: 0/6 [3nd " H.J. Lu
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=6dc9ffc81003070840x649dcc23qa088fddd69df7e54@mail.gmail.com \
--to=hjl.tools@gmail.com \
--cc=gdb-patches@sourceware.org \
--cc=mark.kettenis@xs4all.nl \
/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 read-only IMAP folder(s) and NNTP newsgroup(s).