error in all "patch"
Message
TequilaTR,TequilaTR wrote:I understand, that mounting tmpfs could help temporarily, but people might want to use patch on usual files as well, not only during emerging, right?
Yes, you are correct
Mounting tmpfs was not to be construed as a resolution - it was intended to narrow-down the source of the problem.
It could have been used as a temporary alternative to enable updating of the system without the need to recompile the kernel without selinux support
Good luck
Since installing gentoo, my life has become one long emerge
Since installing gentoo, my life has become one long emerge
There was this test failure in the patch-2.7 test log:
I do not know what part of patch that tests, so I do not know
if that is relevant. (I simply declined to install it for the moment,
since patch-2.6.1 has been working fine and does not
show the test failure, using the patch test suite in
the patch source tree.)
The error as described still looks a lot more like filesystem
corruption than anything else, but it apparently does not
come from the filesystem code itself, since it is seen on more
than one different filesystem. Working around it by not enabling
an SELINUX config option suggests some filesystem-independent
security callback in the VFS corrupting pathnames, but kernel
bugs are rarely that simple (more like "A causes B to do this,
which C had not considered possible, so had not sanity checked
for it, and when it happens, C lets D do this which breaks a kernel
function").
The result can be far from the cause (need to bisect the kernel
patches to find it, in git terms).
Code: Select all
XFAIL: dash-o-append
if that is relevant. (I simply declined to install it for the moment,
since patch-2.6.1 has been working fine and does not
show the test failure, using the patch test suite in
the patch source tree.)
The error as described still looks a lot more like filesystem
corruption than anything else, but it apparently does not
come from the filesystem code itself, since it is seen on more
than one different filesystem. Working around it by not enabling
an SELINUX config option suggests some filesystem-independent
security callback in the VFS corrupting pathnames, but kernel
bugs are rarely that simple (more like "A causes B to do this,
which C had not considered possible, so had not sanity checked
for it, and when it happens, C lets D do this which breaks a kernel
function").
The result can be far from the cause (need to bisect the kernel
patches to find it, in git terms).
TIA