Re: Cross Memory Attach v3

Re: Cross Memory Attach v3

[Geert Uytterhoeven]

Hi Christopher,

On Mon, Jul 18, 2011 at 17:05, Christopher Yeoh <cyeoh@au1.ibm.com> wrote:
> For arch maintainers there are some simple tests to be able to quickly
> verify that the syscalls are working correctly here:

I'm wiring up these new syscalls on m68k.

> http://ozlabs.org/~cyeoh/cma/cma-test-20110718.tgz

The included README talks about:

    setup_process_readv_simple
    setup_process_readv_iovec
   setup_process_writev

while the actual test executables are called:

    setup_process_vm_readv_simple
    setup_process_vm_readv_iovec
    setup_process_vm_writev

On m68k (ARAnyM), the first and third test succeed. The second one
fails, though:

# Setting up target with num iovecs 10, test buffer size 100000
Target process is setup
Run the following to test:
./t_process_vm_readv_iovec 1574 10 0x800030b0 89 0x80003110 38302
0x8000c6b8 22423 0x80011e58 18864 0x80016810 583 0x80016a60 8054
0x800189e0 3417 0x80019740 368 0x800198b8 897 0x80019c40 7003

and in the other window:

# ./t_process_vm_readv_iovec 1574 10 0x800030b0 89 0x80003110 38302
0x8000c6b8 22423 0x80011e58 18864 0x80016810 583 0x80016a60 8054
0x800189e0 3417 0x80019740 368 0x800198b8 897 0x80019c40 7003
copy_from_process failed: Invalid argument
error code: 29
#

Any suggestions?

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: Cross Memory Attach v3

[Christopher Yeoh]

Hi Geert,

On Sun, 20 Nov 2011 11:16:17 +0100
Geert Uytterhoeven <geert@linux-m68k.org> wrote:
> On Mon, Jul 18, 2011 at 17:05, Christopher Yeoh <cyeoh@au1.ibm.com>
> wrote:
> > For arch maintainers there are some simple tests to be able to
> > quickly verify that the syscalls are working correctly here:
> 
> I'm wiring up these new syscalls on m68k.
> 
> > http://ozlabs.org/~cyeoh/cma/cma-test-20110718.tgz
> 
> The included README talks about:
> 
>     setup_process_readv_simple
>     setup_process_readv_iovec
>    setup_process_writev
> 
> while the actual test executables are called:
> 
>     setup_process_vm_readv_simple
>     setup_process_vm_readv_iovec
>     setup_process_vm_writev

Oops. Have fixed this and uploaded a new version

 http://ozlabs.org/~cyeoh/cma/cma-test-20111121.tgz

It also includes another minor change (see below)

> On m68k (ARAnyM), the first and third test succeed. The second one
> fails, though:
> 
> # Setting up target with num iovecs 10, test buffer size 100000
> Target process is setup
> Run the following to test:
> ./t_process_vm_readv_iovec 1574 10 0x800030b0 89 0x80003110 38302
> 0x8000c6b8 22423 0x80011e58 18864 0x80016810 583 0x80016a60 8054
> 0x800189e0 3417 0x80019740 368 0x800198b8 897 0x80019c40 7003
> 
> and in the other window:
> 
> # ./t_process_vm_readv_iovec 1574 10 0x800030b0 89 0x80003110 38302
> 0x8000c6b8 22423 0x80011e58 18864 0x80016810 583 0x80016a60 8054
> 0x800189e0 3417 0x80019740 368 0x800198b8 897 0x80019c40 7003
> copy_from_process failed: Invalid argument

That should say process_vm_readv instead of copy_from_process. The
error message is fixed in the just updated test.

> error code: 29
> #
> 
> Any suggestions?
> 

Given that the first and third tests succeed, I think the problem is
with the iovec parameters. The -EINVAL is most likely coming from
rw_copy_check_uvector. Any chance that something bad is
happening to lvec/liovcnt or rvec/riovcnt in the wireup? 

The iovecs are checked in process_vm_rw before the core of the
process_vm_readv/writev code is called so should be easy to confirm if
this is the problem.

The other couple of places where it could possibly come from is that
for some reason the flags parameter ends up being non zero or when
looking up the task the mm is NULL. But given that the first and second
tests succeed I think its unlikely that either of these is the cause.

Regards,

Chris
-- 
cyeoh@au.ibm.com

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: Cross Memory Attach v3

[Geert Uytterhoeven]

Hi Christopher,

On Mon, Nov 21, 2011 at 01:13, Christopher Yeoh <cyeoh@au1.ibm.com> wrote:
> On Sun, 20 Nov 2011 11:16:17 +0100
> Geert Uytterhoeven <geert@linux-m68k.org> wrote:
>> On Mon, Jul 18, 2011 at 17:05, Christopher Yeoh <cyeoh@au1.ibm.com>
>> wrote:
>> > For arch maintainers there are some simple tests to be able to
>> > quickly verify that the syscalls are working correctly here:
>>
>> I'm wiring up these new syscalls on m68k.
>>
>> On m68k (ARAnyM), the first and third test succeed. The second one
>> fails, though:
>>
>> # Setting up target with num iovecs 10, test buffer size 100000
>> Target process is setup
>> Run the following to test:
>> ./t_process_vm_readv_iovec 1574 10 0x800030b0 89 0x80003110 38302
>> 0x8000c6b8 22423 0x80011e58 18864 0x80016810 583 0x80016a60 8054
>> 0x800189e0 3417 0x80019740 368 0x800198b8 897 0x80019c40 7003
>>
>> and in the other window:
>>
>> # ./t_process_vm_readv_iovec 1574 10 0x800030b0 89 0x80003110 38302
>> 0x8000c6b8 22423 0x80011e58 18864 0x80016810 583 0x80016a60 8054
>> 0x800189e0 3417 0x80019740 368 0x800198b8 897 0x80019c40 7003
>> copy_from_process failed: Invalid argument
>
> That should say process_vm_readv instead of copy_from_process. The
> error message is fixed in the just updated test.
>
>> error code: 29
>> #
>>
>> Any suggestions?
>
> Given that the first and third tests succeed, I think the problem is
> with the iovec parameters. The -EINVAL is most likely coming from
> rw_copy_check_uvector. Any chance that something bad is
> happening to lvec/liovcnt or rvec/riovcnt in the wireup?
>
> The iovecs are checked in process_vm_rw before the core of the
> process_vm_readv/writev code is called so should be easy to confirm if
> this is the problem.
>
> The other couple of places where it could possibly come from is that
> for some reason the flags parameter ends up being non zero or when
> looking up the task the mm is NULL. But given that the first and second
> tests succeed I think its unlikely that either of these is the cause.

It turned out the flags parameter was non-zero, due to syscall() only supporting
up to 5 parameters in the glibc I was using for testing.

I checked the eglibc sources (2.11.1-0ubuntu7.8), and it's still not
fixed there,
although I could find a fix for a similar issue in klibc
(http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=334917).

When forcing flags to zero, it works ;-)
So sorry for bothering you.

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: Cross Memory Attach v3

[Thorsten Glaser]

Geert Uytterhoeven dixit:

>It turned out the flags parameter was non-zero, due to syscall() only supporting
>up to 5 parameters in the glibc I was using for testing.
>
>I checked the eglibc sources (2.11.1-0ubuntu7.8), and it's still not
>fixed there,
>although I could find a fix for a similar issue in klibc
>(http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=334917).

Does this work for you?

Index: eglibc-2.13/ports/sysdeps/unix/sysv/linux/m68k/syscall.S
===================================================================
--- eglibc-2.13.orig/ports/sysdeps/unix/sysv/linux/m68k/syscall.S	2011-12-04 14:36:16.000000000 +0000
+++ eglibc-2.13/ports/sysdeps/unix/sysv/linux/m68k/syscall.S	2011-12-04 14:36:26.000000000 +0000
@@ -24,9 +24,9 @@
 	.text
 ENTRY (syscall)
 	move.l 4(%sp), %d0	/* Load syscall number.  */
-	_DOARGS_5 (24)		/* Frob arguments.  */
+	_DOARGS_6 (28)		/* Frob arguments.  */
 	trap &0			/* Do the system call.  */
-	UNDOARGS_5		/* Unfrob arguments.  */
+	UNDOARGS_6		/* Unfrob arguments.  */
 	cmp.l &-4095, %d0	/* Check %d0 for error.  */
 	jcc SYSCALL_ERROR_LABEL	/* Jump to error handler if negative.  */
 	rts			/* Return to caller.  */

bye,
//mirabilos
-- 
  “Having a smoking section in a restaurant is like having
          a peeing section in a swimming pool.”
						-- Edward Burr

Re: Cross Memory Attach v3

[Geert Uytterhoeven]

Hi Thorsten,

On Sun, Dec 4, 2011 at 15:37, Thorsten Glaser <tg@mirbsd.de> wrote:
> Geert Uytterhoeven dixit:
>>It turned out the flags parameter was non-zero, due to syscall() only supporting
>>up to 5 parameters in the glibc I was using for testing.
>>
>>I checked the eglibc sources (2.11.1-0ubuntu7.8), and it's still not
>>fixed there,
>>although I could find a fix for a similar issue in klibc
>>(http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=334917).
>
> Does this work for you?

Thanks!
Your patch looks fine, but I'm not gonna compile eglibc now.

> Index: eglibc-2.13/ports/sysdeps/unix/sysv/linux/m68k/syscall.S
> ===================================================================
> --- eglibc-2.13.orig/ports/sysdeps/unix/sysv/linux/m68k/syscall.S       2011-12-04 14:36:16.000000000 +0000
> +++ eglibc-2.13/ports/sysdeps/unix/sysv/linux/m68k/syscall.S    2011-12-04 14:36:26.000000000 +0000
> @@ -24,9 +24,9 @@
>        .text
>  ENTRY (syscall)
>        move.l 4(%sp), %d0      /* Load syscall number.  */
> -       _DOARGS_5 (24)          /* Frob arguments.  */
> +       _DOARGS_6 (28)          /* Frob arguments.  */
>        trap &0                 /* Do the system call.  */
> -       UNDOARGS_5              /* Unfrob arguments.  */
> +       UNDOARGS_6              /* Unfrob arguments.  */
>        cmp.l &-4095, %d0       /* Check %d0 for error.  */
>        jcc SYSCALL_ERROR_LABEL /* Jump to error handler if negative.  */
>        rts                     /* Return to caller.  */

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: Cross Memory Attach v3

[Andreas Schwab]

Thorsten Glaser <tg@mirbsd.de> writes:

> Index: eglibc-2.13/ports/sysdeps/unix/sysv/linux/m68k/syscall.S
> ===================================================================
> --- eglibc-2.13.orig/ports/sysdeps/unix/sysv/linux/m68k/syscall.S	2011-12-04 14:36:16.000000000 +0000
> +++ eglibc-2.13/ports/sysdeps/unix/sysv/linux/m68k/syscall.S	2011-12-04 14:36:26.000000000 +0000
> @@ -24,9 +24,9 @@
>  	.text
>  ENTRY (syscall)
>  	move.l 4(%sp), %d0	/* Load syscall number.  */
> -	_DOARGS_5 (24)		/* Frob arguments.  */
> +	_DOARGS_6 (28)		/* Frob arguments.  */
>  	trap &0			/* Do the system call.  */
> -	UNDOARGS_5		/* Unfrob arguments.  */
> +	UNDOARGS_6		/* Unfrob arguments.  */
>  	cmp.l &-4095, %d0	/* Check %d0 for error.  */
>  	jcc SYSCALL_ERROR_LABEL	/* Jump to error handler if negative.  */
>  	rts			/* Return to caller.  */

Thanks, installed.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Cross Memory Attach v3

[Thorsten Glaser]

Andreas Schwab dixit:

>Thorsten Glaser <tg@mirbsd.de> writes:
>
>> Index: eglibc-2.13/ports/sysdeps/unix/sysv/linux/m68k/syscall.S
>> ===================================================================
>> --- eglibc-2.13.orig/ports/sysdeps/unix/sysv/linux/m68k/syscall.S	2011-12-04 14:36:16.000000000 +0000
>> +++ eglibc-2.13/ports/sysdeps/unix/sysv/linux/m68k/syscall.S	2011-12-04 14:36:26.000000000 +0000
>> @@ -24,9 +24,9 @@
>>  	.text
>>  ENTRY (syscall)
>>  	move.l 4(%sp), %d0	/* Load syscall number.  */
>> -	_DOARGS_5 (24)		/* Frob arguments.  */
>> +	_DOARGS_6 (28)		/* Frob arguments.  */
>>  	trap &0			/* Do the system call.  */
>> -	UNDOARGS_5		/* Unfrob arguments.  */
>> +	UNDOARGS_6		/* Unfrob arguments.  */
>>  	cmp.l &-4095, %d0	/* Check %d0 for error.  */
>>  	jcc SYSCALL_ERROR_LABEL	/* Jump to error handler if negative.  */
>>  	rts			/* Return to caller.  */
>
>Thanks, installed.

(Where?) So, should $someone submit that to eglibc-ports and
Debisn src:eglibc? Did it indeed do the trick?

bye,
//mirabilos
-- 
  “Having a smoking section in a restaurant is like having
          a peeing section in a swimming pool.”
						-- Edward Burr

Re: Cross Memory Attach v3

[Aurelien Jarno]

On Sun, Dec 11, 2011 at 10:11:03PM +0000, Thorsten Glaser wrote:
> Andreas Schwab dixit:
> 
> >Thorsten Glaser <tg@mirbsd.de> writes:
> >
> >> Index: eglibc-2.13/ports/sysdeps/unix/sysv/linux/m68k/syscall.S
> >> ===================================================================
> >> --- eglibc-2.13.orig/ports/sysdeps/unix/sysv/linux/m68k/syscall.S	2011-12-04 14:36:16.000000000 +0000
> >> +++ eglibc-2.13/ports/sysdeps/unix/sysv/linux/m68k/syscall.S	2011-12-04 14:36:26.000000000 +0000
> >> @@ -24,9 +24,9 @@
> >>  	.text
> >>  ENTRY (syscall)
> >>  	move.l 4(%sp), %d0	/* Load syscall number.  */
> >> -	_DOARGS_5 (24)		/* Frob arguments.  */
> >> +	_DOARGS_6 (28)		/* Frob arguments.  */
> >>  	trap &0			/* Do the system call.  */
> >> -	UNDOARGS_5		/* Unfrob arguments.  */
> >> +	UNDOARGS_6		/* Unfrob arguments.  */
> >>  	cmp.l &-4095, %d0	/* Check %d0 for error.  */
> >>  	jcc SYSCALL_ERROR_LABEL	/* Jump to error handler if negative.  */
> >>  	rts			/* Return to caller.  */
> >
> >Thanks, installed.
> 
> (Where?) So, should $someone submit that to eglibc-ports and
> Debisn src:eglibc? Did it indeed do the trick?
> 

This patch should be reported to libc-ports@sourceware.org. If it also 
needs to be applied to the Debian package quickly, please open a bug 
report in the Debian BTS.

-- 
Aurelien Jarno	                        GPG: 1024D/F1BCDB73
aurelien@aurel32.net                 http://www.aurel32.net

Re: Cross Memory Attach v3

[Thorsten Glaser]

I’ve tested the below patch locally in the meantime, and it
does not break anything.

Aurelien Jarno dixit:

>On Sun, Dec 11, 2011 at 10:11:03PM +0000, Thorsten Glaser wrote:
>> Andreas Schwab dixit:
>> 
>> >Thorsten Glaser <tg@mirbsd.de> writes:
>> >
>> >> Index: eglibc-2.13/ports/sysdeps/unix/sysv/linux/m68k/syscall.S
>> >> ===================================================================
>> >> --- eglibc-2.13.orig/ports/sysdeps/unix/sysv/linux/m68k/syscall.S	2011-12-04 14:36:16.000000000 +0000
>> >> +++ eglibc-2.13/ports/sysdeps/unix/sysv/linux/m68k/syscall.S	2011-12-04 14:36:26.000000000 +0000
>> >> @@ -24,9 +24,9 @@
>> >>  	.text
>> >>  ENTRY (syscall)
>> >>  	move.l 4(%sp), %d0	/* Load syscall number.  */
>> >> -	_DOARGS_5 (24)		/* Frob arguments.  */
>> >> +	_DOARGS_6 (28)		/* Frob arguments.  */
>> >>  	trap &0			/* Do the system call.  */
>> >> -	UNDOARGS_5		/* Unfrob arguments.  */
>> >> +	UNDOARGS_6		/* Unfrob arguments.  */
>> >>  	cmp.l &-4095, %d0	/* Check %d0 for error.  */
>> >>  	jcc SYSCALL_ERROR_LABEL	/* Jump to error handler if negative.  */
>> >>  	rts			/* Return to caller.  */
>> >
>> >Thanks, installed.
>> 
>> (Where?) So, should $someone submit that to eglibc-ports and
>> Debisn src:eglibc? Did it indeed do the trick?
>
>This patch should be reported to libc-ports@sourceware.org. If it also 

Reporting hereby.

>needs to be applied to the Debian package quickly, please open a bug 
>report in the Debian BTS.

@kernel people, is this needed / useful? Do we want/need this?
(I wonder why it's not using _DOARGS_6 already, as that looks
as if it were designed for it.)

bye,
//mirabilos
-- 
22:20⎜<asarch> The crazy that persists in his craziness becomes a master
22:21⎜<asarch> And the distance between the craziness and geniality is
only measured by the success 18:35⎜<asarch> "Psychotics are consistently
inconsistent. The essence of sanity is to be inconsistently inconsistent