On Wednesday 05 April 2006 20:18, Jeff Dike wrote:
> On Wed, Apr 05, 2006 at 07:16:22PM +0200, Olivier Crameri wrote:
> > Unfortunately, I'm having some weird issues that I can't really
> > understand. I can read the file using fread, but only in a buffer
> > that I allocated using um_kmalloc. If I use a buffer allocated by
> > malloc, the fread fails. Then, even if I replace all my mallocs by
> > um_kmallocs, some libc functions (such as sscanf) don't seem to work
> > properly. I guess I'm missing something, but I can't figure out what.
>
> Define "fails" and "don't seem to work properly".
>
> If your buffers are larger than 128K, then libc malloc gets turned into
> UML kernel vmalloc.
No, I made it "greater than 4k" in 2.4.24 time... in fact it's what's
happening to him (see his last mail):
void *__wrap_malloc(int size)
{
void *ret;
if(!CAN_KMALLOC())
return(__real_malloc(size));
else if(size <= PAGE_SIZE) /* finding contiguos pages can be hard*/
ret = um_kmalloc(size);
else ret = um_vmalloc(size);
}
128K is the maximum conceivable kmallocable size, but it hardly works due to
fragmentation.
> In this case, the buffer isn't mapped, and
> passing it into a system call will make it return -EFAULT. The
> easiest workaround for this is to memset the thing immediately after
> allocating it.
> Also, if you're using the libc things you're talking about, watch out
> for your stack consumption. By default, you get two pages (8K).
By default on UML you get 4 pages, 16K, i.e. PAGE_SIZE * (1 <<
CONFIG_KERNEL_STACK_ORDER) == 4k * (1<<2) == 4k * 4.
> printf will completely use it up, so it is unusable in kernel code.
Indeed. sprintf() + write(1) works, instead - grep for my_printf in the source
code, I've coded this hack up.
> UML kernel stack size is configurable - CONFIG_KERNEL_STACK_ORDER -
> bumping that to 3 will double the kernel stack size. If problems then
> go away, then you know that libc is overflowing your stack.
> Jeff
--
Inform me of my mistakes, so I can keep imitating Homer Simpson's "Doh!".
Paolo Giarrusso, aka Blaisorblade (Skype ID "PaoloGiarrusso", ICQ 215621894)
http://www.user-mode-linux.org/~blaisorblade
___________________________________
Yahoo! Mail: gratis 1GB per i messaggi e allegati da 10MB
http://mail.yahoo.it
-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
_______________________________________________
User-mode-linux-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel