2010/1/18 Dave Martin <email address hidden>:
> I managed to build the kernel modules against the linux-fsl-imx51
> headers and install them; make check then ran, using the configure args
> you suggest (I added --prefix=/usr --mandir=/usr/share/man for
> consistency with the package build proper, but I doubt this affects the
> test results). See the attached log for the results.
>
> A couple of tests are skipped; the prctl test fails. The atomics tests
> appear to pass.
That all looks good - bug2524 test is PPC specific, prctl test fails
because of known problems with configure and the test itself.
> I have not currently done anything in my patch to cause
> R_KCODE___kuser_cmpxchg to be defined: I defined it in my build using a
> masquerade wrapper around GCC. This will need to be addressed in the
> build scripts.
I have a trivial patch I'm going to apply that assumes this will
always be supported. It's the better code path to use, and it's only
untrue pre 2.6.12, which isn't a problem for Debian or Ubuntu.
I'll upload 0.8.2-8 to Debian now, with a reference in the Changelog
to the LP bug.
2010/1/18 Dave Martin <email address hidden>: /usr/share/ man for
> I managed to build the kernel modules against the linux-fsl-imx51
> headers and install them; make check then ran, using the configure args
> you suggest (I added --prefix=/usr --mandir=
> consistency with the package build proper, but I doubt this affects the
> test results). See the attached log for the results.
>
> A couple of tests are skipped; the prctl test fails. The atomics tests
> appear to pass.
That all looks good - bug2524 test is PPC specific, prctl test fails
because of known problems with configure and the test itself.
> I have not currently done anything in my patch to cause __kuser_ cmpxchg to be defined: I defined it in my build using a
> R_KCODE_
> masquerade wrapper around GCC. This will need to be addressed in the
> build scripts.
I have a trivial patch I'm going to apply that assumes this will
always be supported. It's the better code path to use, and it's only
untrue pre 2.6.12, which isn't a problem for Debian or Ubuntu.
I'll upload 0.8.2-8 to Debian now, with a reference in the Changelog
to the LP bug.
Alan