On Tue, 2022-02-08 at 14:05 -0800, Andrii Nakryiko wrote: > On Mon, Feb 7, 2022 at 9:16 PM Ilya Leoshkevich <iii@xxxxxxxxxxxxx> > wrote: > > > > Instead of conditionally overriding PT_REGS_PARM4_SYSCALL, provide > > default fallbacks for all __PT_PARMn_REG_SYSCALL macros, so that > > architectures can simply override a specific syscall parameter > > macro. > > Also allow completely overriding PT_REGS_PARM1_SYSCALL for > > non-trivial access sequences. > > > > Co-developed-by: Naveen N. Rao <naveen.n.rao@xxxxxxxxxxxxxxxxxx> > > Signed-off-by: Naveen N. Rao <naveen.n.rao@xxxxxxxxxxxxxxxxxx> > > Signed-off-by: Ilya Leoshkevich <iii@xxxxxxxxxxxxx> > > --- > > tools/lib/bpf/bpf_tracing.h | 48 +++++++++++++++++++++++++-------- > > ---- > > 1 file changed, 33 insertions(+), 15 deletions(-) > > > > diff --git a/tools/lib/bpf/bpf_tracing.h > > b/tools/lib/bpf/bpf_tracing.h > > index da7e8d5c939c..82f1e935d549 100644 > > --- a/tools/lib/bpf/bpf_tracing.h > > +++ b/tools/lib/bpf/bpf_tracing.h > > @@ -265,25 +265,43 @@ struct pt_regs; > > > > #endif > > > > -#define PT_REGS_PARM1_SYSCALL(x) PT_REGS_PARM1(x) > > -#define PT_REGS_PARM2_SYSCALL(x) PT_REGS_PARM2(x) > > -#define PT_REGS_PARM3_SYSCALL(x) PT_REGS_PARM3(x) > > -#ifdef __PT_PARM4_REG_SYSCALL > > +#ifndef __PT_PARM1_REG_SYSCALL > > +#define __PT_PARM1_REG_SYSCALL __PT_PARM1_REG > > +#endif > > +#ifndef __PT_PARM2_REG_SYSCALL > > +#define __PT_PARM2_REG_SYSCALL __PT_PARM2_REG > > +#endif > > +#ifndef __PT_PARM3_REG_SYSCALL > > +#define __PT_PARM3_REG_SYSCALL __PT_PARM3_REG > > +#endif > > +#ifndef __PT_PARM4_REG_SYSCALL > > +#define __PT_PARM4_REG_SYSCALL __PT_PARM4_REG > > +#endif > > +#ifndef __PT_PARM5_REG_SYSCALL > > +#define __PT_PARM5_REG_SYSCALL __PT_PARM5_REG > > +#endif > > + > > +#ifndef PT_REGS_PARM1_SYSCALL > > +#define PT_REGS_PARM1_SYSCALL(x) (__PT_REGS_CAST(x)- > > >__PT_PARM1_REG_SYSCALL) > > +#endif > > +#ifndef PT_REGS_PARM2_SYSCALL > > +#define PT_REGS_PARM2_SYSCALL(x) (__PT_REGS_CAST(x)- > > >__PT_PARM2_REG_SYSCALL) > > +#endif > > +#ifndef PT_REGS_PARM3_SYSCALL > > +#define PT_REGS_PARM3_SYSCALL(x) (__PT_REGS_CAST(x)- > > >__PT_PARM3_REG_SYSCALL) > > +#endif > > +#ifndef PT_REGS_PARM4_SYSCALL > > #define PT_REGS_PARM4_SYSCALL(x) (__PT_REGS_CAST(x)- > > >__PT_PARM4_REG_SYSCALL) > > -#else /* __PT_PARM4_REG_SYSCALL */ > > -#define PT_REGS_PARM4_SYSCALL(x) PT_REGS_PARM4(x) > > #endif > > -#define PT_REGS_PARM5_SYSCALL(x) PT_REGS_PARM5(x) > > +#ifndef PT_REGS_PARM5_SYSCALL > > +#define PT_REGS_PARM5_SYSCALL(x) (__PT_REGS_CAST(x)- > > >__PT_PARM5_REG_SYSCALL) > > +#endif > > > > -#define PT_REGS_PARM1_CORE_SYSCALL(x) PT_REGS_PARM1_CORE(x) > > -#define PT_REGS_PARM2_CORE_SYSCALL(x) PT_REGS_PARM2_CORE(x) > > -#define PT_REGS_PARM3_CORE_SYSCALL(x) PT_REGS_PARM3_CORE(x) > > -#ifdef __PT_PARM4_REG_SYSCALL > > +#define PT_REGS_PARM1_CORE_SYSCALL(x) > > BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM1_REG_SYSCALL) > > +#define PT_REGS_PARM2_CORE_SYSCALL(x) > > BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM2_REG_SYSCALL) > > +#define PT_REGS_PARM3_CORE_SYSCALL(x) > > BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM3_REG_SYSCALL) > > #define PT_REGS_PARM4_CORE_SYSCALL(x) > > BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM4_REG_SYSCALL) > > -#else /* __PT_PARM4_REG_SYSCALL */ > > -#define PT_REGS_PARM4_CORE_SYSCALL(x) PT_REGS_PARM4_CORE(x) > > -#endif > > -#define PT_REGS_PARM5_CORE_SYSCALL(x) PT_REGS_PARM5_CORE(x) > > +#define PT_REGS_PARM5_CORE_SYSCALL(x) > > BPF_CORE_READ(__PT_REGS_CAST(x), __PT_PARM5_REG_SYSCALL) > > > > No, please don't do it. It makes CORE variants too rigid. We agreed > w/ > Naveen that the way you did it in v2 is better and more flexible and > in v3 you did it the other way. Why? As far as I remember we didn't discuss this proposal from Naveen [1] - there was another one about moving SYS_PREFIX to libbpf, where we agreed that it would have bad consequences. Isn't this patch essentially equivalent to the one from my v3 [2], but with the added ability to override more things and better-looking? I.e.: if we define __PT_PARMn_REG_SYSCALL, then PT_REGS_PARMn_SYSCALL and PT_REGS_PARMn_CORE_SYSCALL use that, and __PT_PARMn_REG otherwise. [1] https://lore.kernel.org/bpf/1643990954.fs9q9mrdxt.naveen@xxxxxxxxxxxxx/ [2] https://lore.kernel.org/bpf/20220204145018.1983773-5-iii@xxxxxxxxxxxxx/ > > > #else /* defined(bpf_target_defined) */ > > > > -- > > 2.34.1 > >
