| Age | Commit message (Collapse) | Author |
|---|
|
delivery. Very early is before fork() returns.
|
|
|
|
|
|
|
|
|
|
correct place.
|
|
|
|
|
|
|
|
For some reason this test doesn't fail which means the code is correct and
my reading of the code is incorrect. That can be either a good or bad
sign depending on how you look at it.
millert@ says it "looks correct"
|
|
|
|
|
|
miod and mickey; ok art@
|
|
range in the current process, using trap #451.
This is necessary for proper gcc trampolines operation, and, later, ld.so...
|
|
aligned, boundary on 32 bit systems.
Written a while ago for m88k, eventually turned into a real regress test.
|
|
|
|
|
|
|
|
the best test around.
|
|
space, as mickey@ pointed out. Enable only for i386 and amd64, since it
might give false negatives on other archs.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Fixes a bogus regression on sparc64. OK henning@
|
|
|
|
|
|
|
|
should happen if signal delivery is not robust.
|
|
pointed out last week, and again today by art.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
ok millert@
|
|
|
|
|
|
Run tests 4 times: with and without blocking and with and without a
timeout for poll/select. Results match Solaris.
|
|
Also test it in regress to be pedantic (though the kernel path is
really the same). While there, remove cred-passing code that we
don't support. OK deraadt@,
|
|
|
|
m88k userland trap tests to its MACHINE_ARCH directory.
|
|
regress conditions (rfmem-stack) are no longer true, and other tests
have to be rewritten to handle shared stacks. spotted by david@
|
|
|
|
|
|
|
|
|
