qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Qemu-devel] Bus errors and /dev/shm -- was: KQEMU bus errors


From: Darryl Dixon
Subject: [Qemu-devel] Bus errors and /dev/shm -- was: KQEMU bus errors
Date: Mon, 14 Feb 2005 09:19:19 +1300

I'm not quite clear on why /dev/shm should affect the 2005-02-11 version and not the 2005-02-09 version of the code, but I took you at your word and tried first with qemu guest mem == size of /dev/shm, and once again received 'Bus error'.  I then set qemu guest mem == 3/4 of /dev/shm and it seems to be working OK.

So, can anyone explain why the apparent regression, and why /dev/shm is used at all?  I would be happier if I understood :)

Incidentally, the 'Protection Error' received by Win9x guest appears to be orthogonal to this problem and isn't fixed by the above like WinXP is.

Cheers,
D


On Sun, 2005-02-13 at 15:51 +0100, Grzegorz Kulewski wrote:
On Sun, 13 Feb 2005, Julian Seward wrote:

> On Sunday 13 February 2005 11:13, Jonas Maebe wrote:
>> On 13 feb 2005, at 10:42, Darryl Dixon wrote:
>>> Most of the references that I can find for a Linux 'Bus error' talk
>>> about unaligned memory accesses.
>>
>> Indeed. The only way I know of to get a bus error under Linux/x86 from
>> a user space program, is to turn on the alignment check flag in the
>> eflags register, followed by an unaligned memory access.
>
> I have a very vague and possibly wrong memory that the another way
> to get a bus error is to mmap a file into an area which is too big
> for the file, then read/write the area beyond the end of the file.
> Or perhaps that was on Solaris.  In any case, it might be worth
> checking this isn't somehow related to mmap or use of mmap.

>From man mmap(2):

"SIGBUS Attempted access to a portion of the buffer that does not 
correspond to the file (for example,
               beyond  the end of the file, including the case where 
another process has truncated the file)."

So this is very probable that the errors are caused by too small 
limit of tmpfs mounted on /dev/shm.


Grzegorz Kulewski



_______________________________________________
Qemu-devel mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/qemu-devel
--
Darryl Dixon <address@hidden>

reply via email to

[Prev in Thread] Current Thread [Next in Thread]