On Tue, May 26, 2020 at 11:26:11AM +0100, Mark Rutland wrote:Yes, it is for 64-bit perf profiling a 32-bit task, not for a compat thread.
On Mon, May 11, 2020 at 10:52:07AM +0800, Jiping Ma wrote:I think this was for the case where you have a 64-bit perf profiling a
Modified the patch subject and the change description.Thanks for this.
PC value is get from regs in REGS_ABI_32 mode, but correct PC
is regs->pc(regs[PERF_REG_ARM64_PC]) in arm64 kernel, which caused
that perf can not parser the backtrace of app with dwarf mode in the
32bit system and 64bit kernel.
Signed-off-by: Jiping Ma <jiping.ma2@xxxxxxxxxxxxx>
---I think there are some more issues here, and we may need a more
arch/arm64/kernel/perf_regs.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/arch/arm64/kernel/perf_regs.c b/arch/arm64/kernel/perf_regs.c
index 0bbac61..0ef2880 100644
@@ -32,6 +32,10 @@ u64 perf_reg_value(struct pt_regs *regs, int idx)
if ((u32)idx == PERF_REG_ARM64_PC)
+ if (perf_reg_abi(current) == PERF_SAMPLE_REGS_ABI_32
+ && idx == 15)
+ return regs->pc;
substantial rework. For a compat thread, we always expose
PERF_SAMPLE_REGS_ABI_32 via per_reg_abi(), but for some reason
perf_reg_value() also munges the compat SP/LR into their ARM64
equivalents, which don't exist in the 32-bit sample ABI. We also don't
zero the regs that don't exist in 32-bit (including the aliasing PC).
32-bit task, and it was passing the registers off to libunwind. Won't that
break if we follow your suggestion?