Matthew Burgess wrote:
Alexander E. Patrakov wrote:
It appears that all leaks reported up to now are unfixable from udev
side (they are due to drivers probing their hardware in a separate
thread)
Should this probing of hardware in a separate thread be reported as a
bug to the driver maintainers, or is it a feature?
This is a feature :(
<FUD probability="nonzero">
Moreover, in linux-2.6.18 SCSI scanning will be asynchronous by default (i.e.,
in a kernel thread), and this means inability to boot from slow SCSI devices
like big hardware RAID arrays without initramfs.
</FUD>
Regardless, please review the following patches. Obviously, the book
patch can't go in until new udev-config and bootscripts tarballs have
been generated.
The patch is OK. I give you the permission to release a new version of udev
config tarball identical to the current one without the "bug" rule and without
nvidia rules.
--
Alexander E. Patrakov
--
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
