On 2018-03-22 12:27 PM, Matt Turner wrote:
On Thu, Mar 22, 2018 at 9:02 AM, John David Anglin <dave.anglin@xxxxxxxx> wrote:
On 2018-03-22 2:46 AM, Matt Turner wrote:
In case anyone is looking for ways of reproducing cache errors... the
test suite in googletest [1] reliably triggers the following:
do_page_fault() command='gtest_break_on_' type=15 address=0x00000000
in libgtest.so[f8493000+57000]
trap #15: Data TLB miss fault
Why do you think this is a cache problem?
Only because I saw a lot more of these when the kernel was much more
unstable on my C8000, and the fixes that made the difference were your
recent TLB flushing changes.
Okay, I see the testsuite hasn't been run in recent Debian builds. I'll
try the testsuite
and see what happens.
I looked at the Gentoo build script. It's probably not easy to
duplicate the build exactly
on Debian given library differences, etc.
As a starting point in debugging these issues, there should be a
register dump for the events
in kern.log or syslog. It should be possible to get a backtrace and a
disassembly around the
exception using gdb. Of course, this is difficult for random events.
Dave
--
John David Anglin dave.anglin@xxxxxxxx
--
To unsubscribe from this list: send the line "unsubscribe linux-parisc" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
