Call Frame Information is used by gdb for back-traces and inserting
breakpoints on function return for the "finish" command. This failed
when inside __kernel_clock_gettime. More concerning than difficulty
debugging is that CFI is also used by stack frame unwinding code to
implement exceptions. If you have an app that needs to handle
asynchronous exceptions for some reason, and you are unlucky enough to
get one inside the VDSO time functions, your app will crash.
What's wrong: There is control flow in __kernel_clock_gettime that
reaches label 99 without saving lr in r12. CFI info however is
interpreted by the unwinder without reference to control flow: It's a
simple matter of "Execute all the CFI opcodes up to the current
address". That means the unwinder thinks r12 contains the return
address at label 99. Disabuse it of that notion by resetting CFI for
the return address at label 99.
Note that the ".cfi_restore lr" could have gone anywhere from the
"mtlr r12" a few instructions earlier to the instruction at label 99.
I put the CFI as late as possible, because in general that's best
practice (and if possible grouped with other CFI in order to reduce
the number of CFI opcodes executed when unwinding). Using r12 as the
return address is perfectly fine after the "mtlr r12" since r12 on
that code path still contains the return address.
__get_datapage also has a CFI error. That function temporarily saves
lr in r0, and reflects that fact with ".cfi_register lr,r0". A later
use of r0 means the CFI at that point isn't correct, as r0 no longer
contains the return address. Fix that too.
Signed-off-by: Alan Modra <amodra@gmail.com>
Tested-by: Reza Arbab <arbab@linux.ibm.com>
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>
__kernel_get_syscall_map() and __kernel_clock_getres() use cmpli to
check if the passed in pointer is non zero. cmpli maps to a 32 bit
compare on binutils, so we ignore the top 32 bits.
A simple test case can be created by passing in a bogus pointer with
the bottom 32 bits clear. Using a clk_id that is handled by the VDSO,
then one that is handled by the kernel shows the problem:
printf("%d\n", clock_getres(CLOCK_REALTIME, (void *)0x100000000));
printf("%d\n", clock_getres(CLOCK_BOOTTIME, (void *)0x100000000));
And we get:
0
-1
The bigger issue is if we pass a valid pointer with the bottom 32 bits
clear, in this case we will return success but won't write any data
to the pointer.
I stumbled across this issue because the LLVM integrated assembler
doesn't accept cmpli with 3 arguments. Fix this by converting them to
cmpldi.
Fixes: a7f290dad3 ("[PATCH] powerpc: Merge vdso's and add vdso support to 32 bits kernel")
Cc: stable@vger.kernel.org # v2.6.15+
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Most architectures use NR_syscalls as the #define for the number of syscalls.
We use __NR_syscalls, and then define NR_syscalls as __NR_syscalls.
__NR_syscalls is not used outside arch code, whereas NR_syscalls is. So as
NR_syscalls must be defined and __NR_syscalls does not, replace __NR_syscalls
with NR_syscalls.
Signed-off-by: Rashmica Gupta <rashmicy@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
powerpc has a link register (lr) used for calling functions. We "bl
<func>" to call a function, and "blr" to return back to the call site.
The lr is only a single register, so if we call another function from
inside this function (ie. nested calls), software must save away the
lr on the software stack before calling the new function. Before
returning (ie. before the "blr"), the lr is restored by software from
the software stack.
This makes branch prediction quite difficult for the processor as it
will only know the branch target just before the "blr".
To help with this, modern powerpc processors keep a (non-architected)
hardware stack of lr called a "link stack". When a "bl <func>" is
run, the lr is pushed onto this stack. When a "blr" is called, the
branch predictor pops the lr value from the top of the link stack, and
uses it to predict the branch target. Hence the processor pipeline
knows a lot earlier the branch target.
This works great but there are some cases where you call "bl" but
without a matching "blr". Once such case is when trying to determine
the program counter (which can't be read directly). Here you "bl+4;
mflr" to get the program counter. If you do this, the link stack will
get out of sync with reality, causing the branch predictor to
mis-predict subsequent function returns.
To avoid this, modern micro-architectures have a special case of bl.
Using the form "bcl 20,31,+4", ensures the processor doesn't push to
the link stack.
The 32 and 64 bit variants of __get_datapage() use a "bl; mflr" to
determine the loaded address of the VDSO. The current versions of
these attempt to use this special bl variant.
Unfortunately they use +8 rather than the required +4. Hence the
current code results in the link stack getting out of sync with
reality and hence the resulting performance degradation.
This patch moves it to bcl+4 by moving __kernel_datapage_offset out of
__get_datapage().
With this patch, running a gettimeofday() (which uses
__get_datapage()) microbenchmark we get a decent bump in performance
on POWER7/8.
For the benchmark in tools/testing/selftests/powerpc/benchmarks/gettimeofday.c
POWER8:
64bit gets ~4% improvement
32bit gets ~9% improvement
POWER7:
64bit gets ~7% improvement
Signed-off-by: Michael Neuling <mikey@neuling.org>
Reported-by: Aaron Sawdey <sawdey@us.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
The vDSO functions should have the same calling convention as a syscall.
Unfortunately, they currently don't set the cr0.so bit which is used to
indicate an error. This patch makes them clear this bit unconditionally
since all functions currently succeed. The syscall fallback done by some
of them will eventually override this if the syscall fails.
This also changes the symbol version of all vdso exports to make sure
glibc can differenciate between old and fixed calls for existing ones
like __kernel_gettimeofday.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This fixes various errors in the new functions added in the vDSO's,
I've now verified all functions on both 32 and 64 bits vDSOs. It also
fix a sign extension bug getting the initial time of day at boot that
could cause the monotonic clock value to be completely on bogus for
64 bits applications (with either the vDSO or the syscall) on
powermacs.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
This patch moves the vdso's to arch/powerpc, adds support for the 32
bits vdso to the 32 bits kernel, rename systemcfg (finally !), and adds
some new (still untested) routines to both vdso's: clock_gettime() with
support for CLOCK_REALTIME and CLOCK_MONOTONIC, clock_getres() (same
clocks) and get_tbfreq() for glibc to retreive the timebase frequency.
Tom,Steve: The implementation of get_tbfreq() I've done for 32 bits
returns a long long (r3, r4) not a long. This is such that if we ever
add support for >4Ghz timebases on ppc32, the userland interface won't
have to change.
I have tested gettimeofday() using some glibc patches in both ppc32 and
ppc64 kernels using 32 bits userland (I haven't had a chance to test a
64 bits userland yet, but the implementation didn't change and was
tested earlier). I haven't tested yet the new functions.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
