| Message ID | 20220209021745.2215452-10-iii@linux.ibm.com (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | fbca4a2f649730b67488a8b36140ce4d2cf13c63 |
| Delegated to: | BPF |
| Headers | show |
| Series | Fix accessing syscall arguments | expand |
On Tue, Feb 8, 2022 at 6:18 PM Ilya Leoshkevich <iii@linux.ibm.com> wrote: > > On arm64, the first syscall argument should be accessed via orig_x0 > (see arch/arm64/include/asm/syscall.h). Currently regs[0] is used > instead, leading to bpf_syscall_macro test failure. > > orig_x0 cannot be added to struct user_pt_regs, since its layout is a > part of the ABI. Therefore provide access to it only through > PT_REGS_PARM1_CORE_SYSCALL() by using a struct pt_regs flavor. > > Reported-by: Heiko Carstens <hca@linux.ibm.com> > Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com> > --- > tools/lib/bpf/bpf_tracing.h | 10 ++++++++++ > 1 file changed, 10 insertions(+) > > diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h > index f364f1f4710e..928f85f7961c 100644 > --- a/tools/lib/bpf/bpf_tracing.h > +++ b/tools/lib/bpf/bpf_tracing.h > @@ -142,8 +142,18 @@ > > #elif defined(bpf_target_arm64) > > +struct pt_regs___arm64 { > + unsigned long orig_x0; > +} __attribute__((preserve_access_index)); > + I just realized that this will probably break anyone who's using old Clang to compile a non-CORE BPF program because preserve_access_index attribute will be unknown. But we don't have to use __attribute__((preserve_access_index)) here, because we use BPF_CORE_READ() in those macro, which will make accesses CO-RE-relocatable anyways. So I dropped __attribute__((preserve_access_index)) for better backwards compatibility. > /* arm64 provides struct user_pt_regs instead of struct pt_regs to userspace */ > #define __PT_REGS_CAST(x) ((const struct user_pt_regs *)(x)) > +#define PT_REGS_PARM1_SYSCALL(x) ({ \ > + _Pragma("GCC error \"PT_REGS_PARM1_SYSCALL() is not supported on arm64, use PT_REGS_PARM1_CORE_SYSCALL() instead\""); \ > + 0l; \ > +}) I shortened message to just "use PT_REGS_PARM1_CORE_SYSCALL() instead" and made it into a single-liner > +#define PT_REGS_PARM1_CORE_SYSCALL(x) \ > + BPF_CORE_READ((const struct pt_regs___arm64 *)(x), orig_x0) also made this into a single-liner > #define __PT_PARM1_REG regs[0] > #define __PT_PARM2_REG regs[1] > #define __PT_PARM3_REG regs[2] > -- > 2.34.1 >
diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h index f364f1f4710e..928f85f7961c 100644 --- a/tools/lib/bpf/bpf_tracing.h +++ b/tools/lib/bpf/bpf_tracing.h @@ -142,8 +142,18 @@ #elif defined(bpf_target_arm64) +struct pt_regs___arm64 { + unsigned long orig_x0; +} __attribute__((preserve_access_index)); + /* arm64 provides struct user_pt_regs instead of struct pt_regs to userspace */ #define __PT_REGS_CAST(x) ((const struct user_pt_regs *)(x)) +#define PT_REGS_PARM1_SYSCALL(x) ({ \ + _Pragma("GCC error \"PT_REGS_PARM1_SYSCALL() is not supported on arm64, use PT_REGS_PARM1_CORE_SYSCALL() instead\""); \ + 0l; \ +}) +#define PT_REGS_PARM1_CORE_SYSCALL(x) \ + BPF_CORE_READ((const struct pt_regs___arm64 *)(x), orig_x0) #define __PT_PARM1_REG regs[0] #define __PT_PARM2_REG regs[1] #define __PT_PARM3_REG regs[2]
On arm64, the first syscall argument should be accessed via orig_x0 (see arch/arm64/include/asm/syscall.h). Currently regs[0] is used instead, leading to bpf_syscall_macro test failure. orig_x0 cannot be added to struct user_pt_regs, since its layout is a part of the ABI. Therefore provide access to it only through PT_REGS_PARM1_CORE_SYSCALL() by using a struct pt_regs flavor. Reported-by: Heiko Carstens <hca@linux.ibm.com> Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com> --- tools/lib/bpf/bpf_tracing.h | 10 ++++++++++ 1 file changed, 10 insertions(+)