PROBLEM: kernel 2.6.22 do_generic_mapping_read() hangs for no apparent reason

[Klaus Knopper <cloop@knopper.net>]

Hi,

Sorry about the quite poor subject. I just got stuck in the debugging
and cannot point to the exact location of the problem, though I am quite
sure it's a kernel 2.6.22 problem, and not a "third-party-module" problem.

cloop-2.06 (block device module similar to loop.c) uses
do_generic_file_read(), which is translated to do_generic_mapping_read()
in mm/filemap.c, which worked fine up to and including Kernel 2.6.21.6.
In Kernel 2.6.22, do_generic_file_read() blocks indefinitely with no log
message at all, whenever data is read from the file and the read_actor
is NOT called (i.e. parts of the file-to-be-read are probably already
cached somewhere).

Unfortunately, apart from cloop, there seem to be not many drivers that
still use do_generic_file_read, so maybe it's time for a rewrite
anyways.

If anybody would volunteer to help debugging or rewriting cloop to work
with Kernels >= 2.6.22, a lot of people developing live CDs would be
very happy (including me. ;-)

cloop source code is at
http://debian-knoppix.alioth.debian.org/sources/cloop_2.06-2.tar.gz
(make cloop.ko KERNEL_DIR=/path/to/kernel/source)

With kind regards
-Klaus Knopper
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/