Returning from an interrupt or syscall to a signal handler currently
begins execution directly at the handler's entry point, with LR set to
the address of the sigreturn trampoline. When the signal handler
function returns, it runs the trampoline. It looks like this:
# interrupt at user address xyz
# kernel stuff... signal is raised
rfid
# void handler(int sig)
addis 2,12,.TOC.-.LCF0@ha
addi 2,2,.TOC.-.LCF0@l
mflr 0
std 0,16(1)
stdu 1,-96(1)
# handler stuff
ld 0,16(1)
mtlr 0
blr
# __kernel_sigtramp_rt64
addi r1,r1,__SIGNAL_FRAMESIZE
li r0,__NR_rt_sigreturn
sc
# kernel executes rt_sigreturn
rfid
# back to user address xyz
Note the blr with no matching bl. This can corrupt the return
predictor.
Solve this by instead resuming execution at the signal trampoline
which then calls the signal handler. qtrace-tools link_stack checker
confirms the entire user/kernel/vdso cycle is balanced after this
patch, whereas it's not upstream.
Alan confirms the dwarf unwind info still looks good. gdb still
recognises the signal frame and can step into parent frames if it
break inside a signal handler.
Performance is pretty noisy, not a very significant change on a POWER9
here, but branch misses are consistently a lot lower on a
microbenchmark:
Performance counter stats for './signal':
13,085.72 msec task-clock # 1.000 CPUs utilized
45,024,760,101 cycles # 3.441 GHz
65,102,895,542 instructions # 1.45 insn per cycle
11,271,673,787 branches # 861.372 M/sec
59,468,979 branch-misses # 0.53% of all branches
12,989.09 msec task-clock # 1.000 CPUs utilized
44,692,719,559 cycles # 3.441 GHz
65,109,984,964 instructions # 1.46 insn per cycle
11,282,136,057 branches # 868.585 M/sec
39,786,942 branch-misses # 0.35% of all branches
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200511101952.1463138-1-npiggin@gmail.com
Based on 1 normalized pattern(s):
this program is free software you can redistribute it and or modify
it under the terms of the gnu general public license as published by
the free software foundation either version 2 of the license or at
your option any later version
extracted by the scancode license scanner the SPDX license identifier
GPL-2.0-or-later
has been chosen to replace the boilerplate/reference in 3029 file(s).
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Allison Randal <allison@lohutok.net>
Cc: linux-spdx@vger.kernel.org
Link: https://lkml.kernel.org/r/20190527070032.746973796@linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
I've finally tracked down why my CR signal-unwind test case still
fails on little-endian. The problem turned to be that the kernel
installs a signal trampoline in the vDSO, and provides a DWARF CFI
record for that trampoline. This CFI describes the save location
for CR:
rsave (70, 38*RSIZE + (RSIZE - CRSIZE))
which is correct for big-endian, but points to the wrong word on
little-endian. This is wrong no matter which ABI.
In addition, for the ELFv2 ABI, we should not only provide a CFI
record for register 70 (cr2), but for all CR fields separately.
Strictly speaking, I guess this would mean providing two separate
vDSO images, one for ELFv1 processes and one for ELFv2 processes (or
maybe playing some tricks with conditional DWARF expressions).
However, having CFI records for the other CR fields in ELFv1 is not
actually wrong, they just will be ignored. So it seems the simplest
fix would be just to always provide CFI for all the fields.
Signed-off-by: Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The current DWARF info for CR are incorrect, causing the gcc unwinder to
go to lunch if we take a segfault in the vdso. This fixes it.
Problem identified by Andrew Haley, and fix provided by Jakub Jelinek
(thanks !).
Unfortunately, a bug in gcc cause it to not quite work either, but that
is being fixed separately with something around the lines of:
linux-unwind.h:
fs->regs.reg[R_CR2].loc.offset = (long) ®s->ccr - new_cfa;
+ /* CR? regs are just 32-bit and PPC is big-endian. */
+ fs->regs.reg[R_CR2].loc.offset += sizeof (long) - 4;
(According to Jakub)
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Mark unwind info for signal trampolines using the new S augmentation flag
introduced in: http://gcc.gnu.org/PR26208.
GCC 4.2 (or patched earlier GCC) will be able to special case unwinding
through frames right above signal trampolines. As the augmentations start
with z flag and S is at the very end of the augmentation string, older GCCs
will just skip the S flag as unknown (that's why an augmentation flag was
chosen over say a new CFA opcode).
Signed-off-by: Jakub Jelinek <jakub@redhat.com>
Cc: Andi Kleen <ak@muc.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.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>
Nicholas Piggin