diff options
| author | Peter Zijlstra <peterz@infradead.org> | 2020-10-30 12:14:21 +0100 |
|---|---|---|
| committer | Peter Zijlstra <peterz@infradead.org> | 2020-11-09 18:12:34 +0100 |
| commit | 76a4efa80900fc40e0fdf243b42aec9fb8c35d24 (patch) | |
| tree | e71668f6517452f1f7c13aac88a9dabab3f47b94 /arch/x86/kernel/perf_regs.c | |
| parent | 09da9c81253dd8e43e0d2d7cea02de6f9f19499d (diff) | |
perf/arch: Remove perf_sample_data::regs_user_copy
struct perf_sample_data lives on-stack, we should be careful about it's
size. Furthermore, the pt_regs copy in there is only because x86_64 is a
trainwreck, solve it differently.
Reported-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Steven Rostedt <rostedt@goodmis.org>
Link: https://lkml.kernel.org/r/20201030151955.258178461@infradead.org
Diffstat (limited to 'arch/x86/kernel/perf_regs.c')
| -rw-r--r-- | arch/x86/kernel/perf_regs.c | 15 |
1 files changed, 11 insertions, 4 deletions
diff --git a/arch/x86/kernel/perf_regs.c b/arch/x86/kernel/perf_regs.c index bb7e1132290b..f9e5352b3bef 100644 --- a/arch/x86/kernel/perf_regs.c +++ b/arch/x86/kernel/perf_regs.c @@ -101,8 +101,7 @@ u64 perf_reg_abi(struct task_struct *task) } void perf_get_regs_user(struct perf_regs *regs_user, - struct pt_regs *regs, - struct pt_regs *regs_user_copy) + struct pt_regs *regs) { regs_user->regs = task_pt_regs(current); regs_user->abi = perf_reg_abi(current); @@ -129,12 +128,20 @@ u64 perf_reg_abi(struct task_struct *task) return PERF_SAMPLE_REGS_ABI_64; } +static DEFINE_PER_CPU(struct pt_regs, nmi_user_regs); + void perf_get_regs_user(struct perf_regs *regs_user, - struct pt_regs *regs, - struct pt_regs *regs_user_copy) + struct pt_regs *regs) { + struct pt_regs *regs_user_copy = this_cpu_ptr(&nmi_user_regs); struct pt_regs *user_regs = task_pt_regs(current); + if (!in_nmi()) { + regs_user->regs = user_regs; + regs_user->abi = perf_reg_abi(current); + return; + } + /* * If we're in an NMI that interrupted task_pt_regs setup, then * we can't sample user regs at all. This check isn't really |