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

Re: CONFIG_USB_STORAGE_DEBUG enabled => linux-image-2.6.16-2-nslu2 is unusable

Frans Pop a écrit :
On Thursday 13 July 2006 00:34, Martin Michlmayr wrote:
Yeah, I just noticed that myself. :/  It does suck but I'm not sure
this is a RC bug.  The kernel does work, sort of, after all... we're
trying to get beta3 out and I think the -boot team will kill me if
there has to be another kernel rebuild so I suggest we leave it as it
is for now (I've disabled it in SVN already for future releases).

Also, nslu2 is not really a mainstream arch that justifies delaying everything. This seems like a too small issue to have to rebuild for all arches to me.

I fully understand your point of view, and I have written in my original bug report that nslu2 is maybe not a subarchitecture release, so the bug is maybe not so critical.

However, I also maintain that this kernel is not usable. Your disk is filled in one or two hours, and the load of the machine is between 1.5 and 2.0.

Ccing -boot to get comments.  Would there be time for another kernel
upload?  If not, I'll downgrade this bug.

An upload would just be possible, especially if it includes no other changes affecting other arches (so we can skip rebuilding the udebs).
The main delay would be the build for arm which takes over a day...
I'll leave the final decision to the kernel team, but if you upload please do so ASAP.

I have reported the bug against the 2.6.16 version as I think it is important for the release. However if you plan to fix the bug later, fixing it in 2.6.17 is also ok for me. I am not sure my machine is fast enough to rebuilt the kernel myself (also given the current load I get doing nothing except filling /var/log/syslog), and I am using this machine to make test builds of the glibc.

Isn't there a way to work around this at run time (i.e. setting a value to 0 in /proc somewhere that disables the debugging)?

Well I have looked at the code, the debugging messages are printed using printk(), and their inclusion or not in the final code is done by preprocessor options (#ifdef #endif).

  .''`.  Aurelien Jarno	            | GPG: 1024D/F1BCDB73
 : :' :  Debian developer           | Electrical Engineer
 `. `'   aurel32@debian.org         | aurelien@aurel32.net
   `-    people.debian.org/~aurel32 | www.aurel32.net

Reply to: