[PATCH bpf v2 1/1] lib: bpf: tracing: fail compilation if target arch is missing

[PATCH bpf v2 1/1] lib: bpf: tracing: fail compilation if target arch is missing

[Lorenz Bauer]

bpf2go is the Go equivalent of libbpf skeleton. The convention is that
the compiled BPF is checked into the repository to facilitate distributing
BPF as part of Go packages. To make this portable, bpf2go by default
generates both bpfel and bpfeb variants of the C.

Using bpf_tracing.h is inherently non-portable since the fields of
struct pt_regs differ between platforms, so CO-RE can't help us here.
The only way of working around this is to compile for each target
platform independently. bpf2go can't do this by default since there
are too many platforms.

Define the various PT_... macros when no target can be determined and
turn them into compilation failures. This works because bpf2go always
compiles for bpf targets, so the compiler fallback doesn't kick in.
Conditionally define __BPF_MISSING_TARGET so that we can inject a
more appropriate error message at build time. The user can then
choose which platform to target explicitly.

Signed-off-by: Lorenz Bauer <lmb@cloudflare.com>
---
 tools/lib/bpf/bpf_tracing.h | 46 +++++++++++++++++++++++++++++++++----
 1 file changed, 42 insertions(+), 4 deletions(-)

diff --git a/tools/lib/bpf/bpf_tracing.h b/tools/lib/bpf/bpf_tracing.h
index c0f3a26aa582..d6bfbe009296 100644
--- a/tools/lib/bpf/bpf_tracing.h
+++ b/tools/lib/bpf/bpf_tracing.h
@@ -25,26 +25,35 @@
 	#define bpf_target_sparc
 	#define bpf_target_defined
 #else
-	#undef bpf_target_defined
-#endif
 
 /* Fall back to what the compiler says */
-#ifndef bpf_target_defined
 #if defined(__x86_64__)
 	#define bpf_target_x86
+	#define bpf_target_defined
 #elif defined(__s390__)
 	#define bpf_target_s390
+	#define bpf_target_defined
 #elif defined(__arm__)
 	#define bpf_target_arm
+	#define bpf_target_defined
 #elif defined(__aarch64__)
 	#define bpf_target_arm64
+	#define bpf_target_defined
 #elif defined(__mips__)
 	#define bpf_target_mips
+	#define bpf_target_defined
 #elif defined(__powerpc__)
 	#define bpf_target_powerpc
+	#define bpf_target_defined
 #elif defined(__sparc__)
 	#define bpf_target_sparc
+	#define bpf_target_defined
+#endif /* no compiler target */
+
 #endif
+
+#ifndef __BPF_TARGET_MISSING
+#define __BPF_TARGET_MISSING "GCC error \"Must specify a BPF target arch via __TARGET_ARCH_xxx\""
 #endif
 
 #if defined(bpf_target_x86)
@@ -287,7 +296,7 @@ struct pt_regs;
 #elif defined(bpf_target_sparc)
 #define BPF_KPROBE_READ_RET_IP(ip, ctx)		({ (ip) = PT_REGS_RET(ctx); })
 #define BPF_KRETPROBE_READ_RET_IP		BPF_KPROBE_READ_RET_IP
-#else
+#elif defined(bpf_target_defined)
 #define BPF_KPROBE_READ_RET_IP(ip, ctx)					    \
 	({ bpf_probe_read_kernel(&(ip), sizeof(ip), (void *)PT_REGS_RET(ctx)); })
 #define BPF_KRETPROBE_READ_RET_IP(ip, ctx)				    \
@@ -295,6 +304,35 @@ struct pt_regs;
 			  (void *)(PT_REGS_FP(ctx) + sizeof(ip))); })
 #endif
 
+#if !defined(bpf_target_defined)
+
+#define PT_REGS_PARM1(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_PARM2(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_PARM3(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_PARM4(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_PARM5(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_RET(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_FP(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_RC(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_SP(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_IP(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+
+#define PT_REGS_PARM1_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_PARM2_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_PARM3_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_PARM4_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_PARM5_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_RET_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_FP_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_RC_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_SP_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define PT_REGS_IP_CORE(x) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+
+#define BPF_KPROBE_READ_RET_IP(ip, ctx) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+#define BPF_KRETPROBE_READ_RET_IP(ip, ctx) ({ _Pragma(__BPF_TARGET_MISSING); 0l; })
+
+#endif /* !defined(bpf_target_defined) */
+
 #ifndef ___bpf_concat
 #define ___bpf_concat(a, b) a ## b
 #endif
-- 
2.30.2

Re: [PATCH bpf v2 1/1] lib: bpf: tracing: fail compilation if target arch is missing

[patchwork-bot+netdevbpf]

Hello:

This patch was applied to bpf/bpf-next.git (refs/heads/master):

On Wed, 16 Jun 2021 09:36:35 +0100 you wrote:
> bpf2go is the Go equivalent of libbpf skeleton. The convention is that
> the compiled BPF is checked into the repository to facilitate distributing
> BPF as part of Go packages. To make this portable, bpf2go by default
> generates both bpfel and bpfeb variants of the C.
> 
> Using bpf_tracing.h is inherently non-portable since the fields of
> struct pt_regs differ between platforms, so CO-RE can't help us here.
> The only way of working around this is to compile for each target
> platform independently. bpf2go can't do this by default since there
> are too many platforms.
> 
> [...]

Here is the summary with links:
  - [bpf,v2,1/1] lib: bpf: tracing: fail compilation if target arch is missing
    https://git.kernel.org/bpf/bpf-next/c/4a638d581a7a

You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html

Re: [PATCH bpf v2 1/1] lib: bpf: tracing: fail compilation if target arch is missing

[Andrii Nakryiko]

On Wed, Jun 16, 2021 at 1:36 AM Lorenz Bauer <lmb@cloudflare.com> wrote:
>
> bpf2go is the Go equivalent of libbpf skeleton. The convention is that
> the compiled BPF is checked into the repository to facilitate distributing
> BPF as part of Go packages. To make this portable, bpf2go by default
> generates both bpfel and bpfeb variants of the C.
>
> Using bpf_tracing.h is inherently non-portable since the fields of
> struct pt_regs differ between platforms, so CO-RE can't help us here.
> The only way of working around this is to compile for each target
> platform independently. bpf2go can't do this by default since there
> are too many platforms.
>
> Define the various PT_... macros when no target can be determined and
> turn them into compilation failures. This works because bpf2go always
> compiles for bpf targets, so the compiler fallback doesn't kick in.
> Conditionally define __BPF_MISSING_TARGET so that we can inject a
> more appropriate error message at build time. The user can then
> choose which platform to target explicitly.
>
> Signed-off-by: Lorenz Bauer <lmb@cloudflare.com>
> ---

Applied to bpf-next, thanks.

But please use one of few "canonical" patch subject prefixes next time:
  - libbpf:
  - selftests/bpf:
  - bpftool:
  - bpf: (for kernel patches)

We don't do ":each: folder: path:" prefixes, it's not necessary and is
quite long.

>  tools/lib/bpf/bpf_tracing.h | 46 +++++++++++++++++++++++++++++++++----
>  1 file changed, 42 insertions(+), 4 deletions(-)
>

[...]