[Qemu-devel] Mips guest

[Qemu-devel] Mips guest

[Nigel Horne]

[-- Attachment #1: Type: text/plain, Size: 1423 bytes --]

The latest CVS snapshot has broken MIPS emulation.

Guest: MIPS debian linux
Host: Fedora 7/x86

The guest boots as far as

CPU revision is: 00019300
FPU revision is: 00739300
Determined physical RAM map:
 memory: 04000000 @ 00000000 (usable)
Initial ramdisk at: 0x802d7000 (2897359 bytes)
Built 1 zonelists.  Total pages: 16384
Kernel command line: rd_start=0x802d7000 rd_size=2897359 root=/dev/hda1 console=ttyS0
Primary instruction cache 2kB, physically tagged, 2-way, linesize 16 bytes.
Primary data cache 2kB, 2-way, linesize 16 bytes.
Synthesized TLB refill handler (20 instructions).
Synthesized TLB load handler fastpath (32 instructions).
Synthesized TLB store handler fastpath (32 instructions).
Synthesized TLB modify handler fastpath (31 instructions).
Cache parity protection enabled
PID hash table entries: 512 (order: 9, 2048 bytes)
Using 100.000 MHz high precision timer.
Console: colour VGA+ 80x25
Dentry cache hash table entries: 8192 (order: 3, 32768 bytes)
Inode-cache hash table entries: 4096 (order: 2, 16384 bytes)
Memory: 59168k/65536k available (2101k kernel code, 6312k reserved, 450k data, 132k init, 0k highmem)

Then it hangs.

The last time I tried this guest was a couple of weeks ago, or so, and all was fine then.

-Nigel

-- 
Nigel Horne. Adjudicator, Arranger, Band Trainer, Composer, Tutor, Typesetter.
NJH Music, Barnsley, UK.  ICQ#20252325
njh@bandsman.co.uk http://www.bandsman.co.uk

[-- Attachment #2: njh.vcf --]
[-- Type: text/x-vcard, Size: 279 bytes --]

begin:vcard
fn:Nigel Horne
n:Horne;Nigel
org:NJH Music
adr:Pogmoor;;28 Wharfedale Road;Barnsley;South Yorkshire;S75 2LJ;UK
email;internet:njh@bandsman.co.uk
tel;fax:+44 870 705 9334
note:Skype: nigelhorne
x-mozilla-html:FALSE
url:http://www.bandsman.co.uk
version:2.1
end:vcard

Re: [Qemu-devel] Mips guest

[Johannes Schindelin]

Hi,

On Wed, 19 Sep 2007, Nigel Horne wrote:

> The latest CVS snapshot has broken MIPS emulation.
> 
> [...]
>
> Then it hangs.
> 
> The last time I tried this guest was a couple of weeks ago, or so, and 
> all was fine then.

It would be so good if you could bisect this.

If you have git, clone qemu from git://git.kernel.dk/data/git/qemu.git, 
and use something "git log" to identify a commit that is likely to be 
good.  The easiest way to proceed is then to copy the commit name (this 
40-character hex string), and use it in

	git bisect start
	git bisect good <that-string>
	git bisect bad HEAD

It will find the middle point between the two, which you should compile 
and test.  Depending on the outcome of the test, you should mark the 
current commit as "git bisect good" or "git bisect bad" (you do not need 
to say which commit, if it is the current you mean).

Eventually, this procedure will give you the bad commit, which then makes 
it easier to see what actually broke the MIPS guest for you.

Thanks,
Dscho

Re: [Qemu-devel] Mips guest

[Nigel Horne]

[-- Attachment #1: Type: text/plain, Size: 2449 bytes --]

I suddenly remembered I'd forgotten to do a "make clean" after my last checkout (oops).

So I did that, and got this compilation error:

...
In file included from /home/njh/src/qemu/linux-user/syscall.c:4040:
/home/njh/src/qemu/linux-user/syscall.c: In function `do_syscall':
/usr/include/bits/unistd.h:62: sorry, unimplemented: inlining failed in call to 'pread': recursive inlining
/home/njh/src/qemu/linux-user/syscall.c:63: sorry, unimplemented: called from here
make[1]: *** [syscall.o] Error 1
make[1]: Leaving directory `/home/njh/src/qemu/i386-linux-user'
make: *** [subdir-i386-linux-user] Error 2

This is because both unistd.h and linux/unistd.h are included. Remove unistd.h from line 27 and that compiles OK.

The next (similar) failure is in mmap.c:

...
/home/njh/src/qemu/linux-user/mmap.c:392: warning: comparison between signed and unsigned
/home/njh/src/qemu/linux-user/mmap.c: In function `mmap_frag':
/usr/include/bits/unistd.h:62: sorry, unimplemented: inlining failed in call to 'pread': recursive inlining
/home/njh/src/qemu/linux-user/mmap.c:63: sorry, unimplemented: called from here
/home/njh/src/qemu/linux-user/mmap.c:139: warning: ignoring return value of `pread', declared with attribute warn_unused_result

Here the fix was on line 24 to replace #include <unistd.h> with <linux/unistd.h>

Then:

...
/home/njh/src/qemu/linux-user/flatload.c: In function `target_pread':
/usr/include/bits/unistd.h:62: sorry, unimplemented: inlining failed in call to 'pread': recursive inlining
/home/njh/src/qemu/linux-user/flatload.c:63: sorry, unimplemented: called from here

Which needs the same fix at line 41.

I suspect these may break building on other OS's, so a cleaner fix will be needed.

The next compilation error is:

...
/home/njh/src/qemu/monitor.c:52: warning: function declaration isn't a prototype
/home/njh/src/qemu/monitor.c:61: error: array size missing in 'term_cmds'
/home/njh/src/qemu/monitor.c:62: error: array size missing in 'info_cmds'

The fix is to change term_cmds[] to *term_cmds, and similarly to info_cmds. This is less than optimal,
but it gets the code compiling pending a code tidy.

The good news is that with these fixes it all compiles and the mips guest runs again (note to self: remember
to run make clean ;-) )


-Nigel

-- 
Nigel Horne. Adjudicator, Arranger, Band Trainer, Composer, Tutor, Typesetter.
NJH Music, Barnsley, UK.  ICQ#20252325
njh@bandsman.co.uk http://www.bandsman.co.uk

[-- Attachment #2: njh.vcf --]
[-- Type: text/x-vcard, Size: 279 bytes --]

begin:vcard
fn:Nigel Horne
n:Horne;Nigel
org:NJH Music
adr:Pogmoor;;28 Wharfedale Road;Barnsley;South Yorkshire;S75 2LJ;UK
email;internet:njh@bandsman.co.uk
tel;fax:+44 870 705 9334
note:Skype: nigelhorne
x-mozilla-html:FALSE
url:http://www.bandsman.co.uk
version:2.1
end:vcard

Re: [Qemu-devel] Mips guest

[Thiemo Seufer]

Nigel Horne wrote:
> I suddenly remembered I'd forgotten to do a "make clean" after my last 
> checkout (oops).
>
> So I did that, and got this compilation error:
>
> ...
> In file included from /home/njh/src/qemu/linux-user/syscall.c:4040:
> /home/njh/src/qemu/linux-user/syscall.c: In function `do_syscall':
> /usr/include/bits/unistd.h:62: sorry, unimplemented: inlining failed in 
> call to 'pread': recursive inlining
> /home/njh/src/qemu/linux-user/syscall.c:63: sorry, unimplemented: called 
> from here
> make[1]: *** [syscall.o] Error 1
> make[1]: Leaving directory `/home/njh/src/qemu/i386-linux-user'
> make: *** [subdir-i386-linux-user] Error 2
>
> This is because both unistd.h and linux/unistd.h are included. Remove 
> unistd.h from line 27 and that compiles OK.

I don't see this on my systems. It looks like broken kernel headers.
What host (OS version) do you use?


Thiemo

RE: [Qemu-devel] Mips guest

[Nigel Horne]

> I don't see this on my systems. It looks like broken kernel headers.
> What host (OS version) do you use?

Linux/x86 (F7)

> 
> Thiemo
>

-Nigel 

Re: [Qemu-devel] Mips guest

[Ronald]

Nigel Horne schreef:
>> I don't see this on my systems. It looks like broken kernel headers.
>> What host (OS version) do you use?
>>     
>
> Linux/x86 (F7)
>
>   
>> Thiemo
>>
>>     
>
> -Nigel 
>
>
>
>   
I think Thiemo also wants to know the kernel version.....

Re: [Qemu-devel] Mips guest

[Nigel Horne]

[-- Attachment #1: Type: text/plain, Size: 644 bytes --]

Ronald wrote:
> Nigel Horne schreef:
>>> I don't see this on my systems. It looks like broken kernel headers.
>>> What host (OS version) do you use?
>>>     
>>
>> Linux/x86 (F7)

That should have read FC6, sorry.

>>
>>  
>>> Thiemo
>>>
>>>     
>>
>> -Nigel
>>
>>
>>   
> I think Thiemo also wants to know the kernel version.....

[njh@njh ~]$ uname -a
Linux njh.bandsman.co.uk 2.6.22.5-49.fc6 #1 SMP Thu Aug 30 14:25:46 EDT 2007 i686 i686 i386 GNU/Linux
[njh@njh ~]$ 

> 


-- 
Nigel Horne. Adjudicator, Arranger, Band Trainer, Composer, Tutor, Typesetter.
NJH Music, Barnsley, UK.  ICQ#20252325
njh@bandsman.co.uk http://www.bandsman.co.uk

[-- Attachment #2: njh.vcf --]
[-- Type: text/x-vcard, Size: 279 bytes --]

begin:vcard
fn:Nigel Horne
n:Horne;Nigel
org:NJH Music
adr:Pogmoor;;28 Wharfedale Road;Barnsley;South Yorkshire;S75 2LJ;UK
email;internet:njh@bandsman.co.uk
tel;fax:+44 870 705 9334
note:Skype: nigelhorne
x-mozilla-html:FALSE
url:http://www.bandsman.co.uk
version:2.1
end:vcard

Re: [Qemu-devel] Mips guest

[Thiemo Seufer]

Nigel Horne wrote:
> The latest CVS snapshot has broken MIPS emulation.
>
> Guest: MIPS debian linux
> Host: Fedora 7/x86
>
> The guest boots as far as
>
> CPU revision is: 00019300
> FPU revision is: 00739300
> Determined physical RAM map:
> memory: 04000000 @ 00000000 (usable)
> Initial ramdisk at: 0x802d7000 (2897359 bytes)
> Built 1 zonelists.  Total pages: 16384
> Kernel command line: rd_start=0x802d7000 rd_size=2897359 root=/dev/hda1 
> console=ttyS0
> Primary instruction cache 2kB, physically tagged, 2-way, linesize 16 bytes.
> Primary data cache 2kB, 2-way, linesize 16 bytes.
> Synthesized TLB refill handler (20 instructions).
> Synthesized TLB load handler fastpath (32 instructions).
> Synthesized TLB store handler fastpath (32 instructions).
> Synthesized TLB modify handler fastpath (31 instructions).
> Cache parity protection enabled
> PID hash table entries: 512 (order: 9, 2048 bytes)
> Using 100.000 MHz high precision timer.
> Console: colour VGA+ 80x25
> Dentry cache hash table entries: 8192 (order: 3, 32768 bytes)
> Inode-cache hash table entries: 4096 (order: 2, 16384 bytes)
> Memory: 59168k/65536k available (2101k kernel code, 6312k reserved, 450k 
> data, 132k init, 0k highmem)
>
> Then it hangs.
>
> The last time I tried this guest was a couple of weeks ago, or so, and all 
> was fine then.

Latest CVS works for me. I'm not aware of intermittent breakage.


Thiemo