[B.A.T.M.A.N.] Converting from /dev to debugfs

Sven Eckelmann sven.eckelmann at gmx.de
Tue May 18 22:35:14 CEST 2010


Sven Eckelmann wrote:
> Unfortunately I am not able to continue it today, but don't want to stop
> other people working on similar things. I will attach v0 (aka works for
> me, but maybe not perfect/good enough) versions of these patches to
> support them. Please feel free to discuss the patches.

I converted the remaining files from sysfs to debugfs, but I noticed that 
there is a problem with the output functions (see  "BUG" comments in patch 3). 
The problem is that we may sleep during the prints in the output functions, 
but we disabled interrupts.... which is bad. This problem was also their 
before we converted from /proc to /sys. It seems that it was first introduced 
in r1517. The old way for hna_global_fill_buffer_text 
hna_local_fill_buffer_text solved it, but restricted the output to 4096 bytes. 
So we could either stay with a "slow" and maybe inconsistent output or 
restricted output length or copy the data to a local buffer and then creating 
the output without spinlock or something completely different.

The patches are for trunk, but I would provide backported versions when the 
patches are discussed and we could solve the problem in a sane way.

Best regards,
	Sven
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.open-mesh.org/pipermail/b.a.t.m.a.n/attachments/20100518/7b240de1/attachment.pgp>


More information about the B.A.T.M.A.N mailing list