mbox series

[0/2] linux-user: Load a vdso for x86_64

Message ID 20200519194452.9009-1-richard.henderson@linaro.org
Headers show
Series linux-user: Load a vdso for x86_64 | expand

Message

Richard Henderson May 19, 2020, 7:44 p.m. UTC
The subject of AT_SYSINFO came up on launchpad recently.

There is definite room for improvement in all of this:

(1) We could build the vdso binary into qemu instead of really
    loading it from the file system.  This would obviate the
    several problems of locating the .so file.  It would also
    mean that --static builds continue to create a standalone
    qemu binary.

(2) We could use our cross-build system to build the vdso.
    Though we'd still likely want to keep the image in git
    along side the other rom images for when cross-build is
    not available.

(3) There are some ??? comments where some decisions could be made,
    and other ??? that are merely commenting on weirdness.

(4) It shouldn't take too much effort to create vdsos for the
    other architectures.  But we should get this one as clean
    as we can first.

Amusingly, this patch set has just turned 10 years old.
First posted April 4, 2010.  I don't recall ever seeing
any review on the several postings over the years.


r~


Richard Henderson (2):
  linux-user: Build vdso for x64.
  linux-user: Load a VDSO for x86-64.

 Makefile                  |   4 +-
 linux-user/elfload.c      | 203 +++++++++++++++++++++++++++++++++++++-
 pc-bios/Makefile          |   5 +
 pc-bios/vdso-linux-x64.S  | 115 +++++++++++++++++++++
 pc-bios/vdso-linux-x64.ld |  81 +++++++++++++++
 pc-bios/vdso-linux-x64.so | Bin 0 -> 7500 bytes
 6 files changed, 401 insertions(+), 7 deletions(-)
 create mode 100644 pc-bios/vdso-linux-x64.S
 create mode 100644 pc-bios/vdso-linux-x64.ld
 create mode 100755 pc-bios/vdso-linux-x64.so

-- 
2.20.1

Comments

Laurent Vivier May 28, 2020, 9:54 a.m. UTC | #1
Le 19/05/2020 à 21:44, Richard Henderson a écrit :
> The subject of AT_SYSINFO came up on launchpad recently.

> 

> There is definite room for improvement in all of this:

> 

> (1) We could build the vdso binary into qemu instead of really

>     loading it from the file system.  This would obviate the

>     several problems of locating the .so file.  It would also

>     mean that --static builds continue to create a standalone

>     qemu binary.

> 

> (2) We could use our cross-build system to build the vdso.

>     Though we'd still likely want to keep the image in git

>     along side the other rom images for when cross-build is

>     not available.

> 

> (3) There are some ??? comments where some decisions could be made,

>     and other ??? that are merely commenting on weirdness.

> 

> (4) It shouldn't take too much effort to create vdsos for the

>     other architectures.  But we should get this one as clean

>     as we can first.

> 

> Amusingly, this patch set has just turned 10 years old.

> First posted April 4, 2010.  I don't recall ever seeing

> any review on the several postings over the years.

> 

> 

> r~

> 

> 

> Richard Henderson (2):

>   linux-user: Build vdso for x64.

>   linux-user: Load a VDSO for x86-64.

> 

>  Makefile                  |   4 +-

>  linux-user/elfload.c      | 203 +++++++++++++++++++++++++++++++++++++-

>  pc-bios/Makefile          |   5 +

>  pc-bios/vdso-linux-x64.S  | 115 +++++++++++++++++++++

>  pc-bios/vdso-linux-x64.ld |  81 +++++++++++++++

>  pc-bios/vdso-linux-x64.so | Bin 0 -> 7500 bytes

>  6 files changed, 401 insertions(+), 7 deletions(-)

>  create mode 100644 pc-bios/vdso-linux-x64.S

>  create mode 100644 pc-bios/vdso-linux-x64.ld

>  create mode 100755 pc-bios/vdso-linux-x64.so

> 


Applied to my linux-user branch.

Thanks,
Laurent
Peter Maydell May 28, 2020, 10:08 a.m. UTC | #2
On Tue, 19 May 2020 at 20:45, Richard Henderson
<richard.henderson@linaro.org> wrote:
>  Makefile                  |   4 +-

>  linux-user/elfload.c      | 203 +++++++++++++++++++++++++++++++++++++-

>  pc-bios/Makefile          |   5 +

>  pc-bios/vdso-linux-x64.S  | 115 +++++++++++++++++++++

>  pc-bios/vdso-linux-x64.ld |  81 +++++++++++++++

>  pc-bios/vdso-linux-x64.so | Bin 0 -> 7500 bytes


I'm not really a fan of binaries in source control :-(

thanks
-- PMM
Laurent Vivier May 28, 2020, 10:32 a.m. UTC | #3
Le 28/05/2020 à 12:08, Peter Maydell a écrit :
> On Tue, 19 May 2020 at 20:45, Richard Henderson

> <richard.henderson@linaro.org> wrote:

>>  Makefile                  |   4 +-

>>  linux-user/elfload.c      | 203 +++++++++++++++++++++++++++++++++++++-

>>  pc-bios/Makefile          |   5 +

>>  pc-bios/vdso-linux-x64.S  | 115 +++++++++++++++++++++

>>  pc-bios/vdso-linux-x64.ld |  81 +++++++++++++++

>>  pc-bios/vdso-linux-x64.so | Bin 0 -> 7500 bytes

> 

> I'm not really a fan of binaries in source control :-(


Can't we see that as a firmware or a ROM?
It's only 7,4 KB and needs a cross-compilation env to be rebuilt.

Do you have another solution?

If you don't like this I can remove the series. Let me know.

Thanks,
Laurent
Philippe Mathieu-Daudé May 28, 2020, 1:20 p.m. UTC | #4
On 5/28/20 12:32 PM, Laurent Vivier wrote:
> Le 28/05/2020 à 12:08, Peter Maydell a écrit :

>> On Tue, 19 May 2020 at 20:45, Richard Henderson

>> <richard.henderson@linaro.org> wrote:

>>>  Makefile                  |   4 +-

>>>  linux-user/elfload.c      | 203 +++++++++++++++++++++++++++++++++++++-

>>>  pc-bios/Makefile          |   5 +

>>>  pc-bios/vdso-linux-x64.S  | 115 +++++++++++++++++++++

>>>  pc-bios/vdso-linux-x64.ld |  81 +++++++++++++++

>>>  pc-bios/vdso-linux-x64.so | Bin 0 -> 7500 bytes

>>

>> I'm not really a fan of binaries in source control :-(

> 

> Can't we see that as a firmware or a ROM?

> It's only 7,4 KB and needs a cross-compilation env to be rebuilt.

> 

> Do you have another solution?


There are only 13 asm instructions, the rest is ELF structure.
Maybe we can strip the linker version ELF sections. Taking out the
alignment gap, not much left.

Nonetheless it is certainly annoying to write that using a readable
scripting language when the linker already does the job, and the patch
is already written/tested.

> 

> If you don't like this I can remove the series. Let me know.

> 

> Thanks,

> Laurent

> 

>
Richard Henderson May 28, 2020, 9:42 p.m. UTC | #5
On 5/28/20 3:32 AM, Laurent Vivier wrote:
> Le 28/05/2020 à 12:08, Peter Maydell a écrit :

>> On Tue, 19 May 2020 at 20:45, Richard Henderson

>> <richard.henderson@linaro.org> wrote:

>>>  Makefile                  |   4 +-

>>>  linux-user/elfload.c      | 203 +++++++++++++++++++++++++++++++++++++-

>>>  pc-bios/Makefile          |   5 +

>>>  pc-bios/vdso-linux-x64.S  | 115 +++++++++++++++++++++

>>>  pc-bios/vdso-linux-x64.ld |  81 +++++++++++++++

>>>  pc-bios/vdso-linux-x64.so | Bin 0 -> 7500 bytes

>>

>> I'm not really a fan of binaries in source control :-(

> 

> Can't we see that as a firmware or a ROM?

> It's only 7,4 KB and needs a cross-compilation env to be rebuilt.

> 

> Do you have another solution?

> 

> If you don't like this I can remove the series. Let me know.


I think some more of the questions in the cover letter need answering.  Does
this patch set not break your own --static chroot tests, for example?


r~
Laurent Vivier May 29, 2020, 8:50 a.m. UTC | #6
Le 28/05/2020 à 23:42, Richard Henderson a écrit :
> On 5/28/20 3:32 AM, Laurent Vivier wrote:

>> Le 28/05/2020 à 12:08, Peter Maydell a écrit :

>>> On Tue, 19 May 2020 at 20:45, Richard Henderson

>>> <richard.henderson@linaro.org> wrote:

>>>>  Makefile                  |   4 +-

>>>>  linux-user/elfload.c      | 203 +++++++++++++++++++++++++++++++++++++-

>>>>  pc-bios/Makefile          |   5 +

>>>>  pc-bios/vdso-linux-x64.S  | 115 +++++++++++++++++++++

>>>>  pc-bios/vdso-linux-x64.ld |  81 +++++++++++++++

>>>>  pc-bios/vdso-linux-x64.so | Bin 0 -> 7500 bytes

>>>

>>> I'm not really a fan of binaries in source control :-(

>>

>> Can't we see that as a firmware or a ROM?

>> It's only 7,4 KB and needs a cross-compilation env to be rebuilt.

>>

>> Do you have another solution?

>>

>> If you don't like this I can remove the series. Let me know.

> 

> I think some more of the questions in the cover letter need answering.  Does

> this patch set not break your own --static chroot tests, for example?


I will test my branch with this series in my --static chroot and remove
the series before the PR.

Thanks,
Laurent
Philippe Mathieu-Daudé May 29, 2020, 8:58 a.m. UTC | #7
On 5/29/20 10:50 AM, Laurent Vivier wrote:
> Le 28/05/2020 à 23:42, Richard Henderson a écrit :

>> On 5/28/20 3:32 AM, Laurent Vivier wrote:

>>> Le 28/05/2020 à 12:08, Peter Maydell a écrit :

>>>> On Tue, 19 May 2020 at 20:45, Richard Henderson

>>>> <richard.henderson@linaro.org> wrote:

>>>>>  Makefile                  |   4 +-

>>>>>  linux-user/elfload.c      | 203 +++++++++++++++++++++++++++++++++++++-

>>>>>  pc-bios/Makefile          |   5 +

>>>>>  pc-bios/vdso-linux-x64.S  | 115 +++++++++++++++++++++

>>>>>  pc-bios/vdso-linux-x64.ld |  81 +++++++++++++++

>>>>>  pc-bios/vdso-linux-x64.so | Bin 0 -> 7500 bytes

>>>>

>>>> I'm not really a fan of binaries in source control :-(

>>>

>>> Can't we see that as a firmware or a ROM?

>>> It's only 7,4 KB and needs a cross-compilation env to be rebuilt.

>>>

>>> Do you have another solution?

>>>

>>> If you don't like this I can remove the series. Let me know.

>>

>> I think some more of the questions in the cover letter need answering.  Does

>> this patch set not break your own --static chroot tests, for example?

> 

> I will test my branch with this series in my --static chroot and remove

> the series before the PR.


Another option is to keep patch #1 in your pullreq, but removing the .so
binary...

> 

> Thanks,

> Laurent

>