On Wed, Oct 27, 2021 at 11:18:22AM +0000, Tong Tiangen wrote:

When a tracing BPF program attempts to read memory without using the
bpf_probe_read() helper, the verifier marks the load instruction with
the BPF_PROBE_MEM flag. Since the riscv JIT does not currently recognize
this flag it falls back to the interpreter.

Add support for BPF_PROBE_MEM, by appending an exception table to the
BPF program. If the load instruction causes a data abort, the fixup
infrastructure finds the exception table and fixes up the fault, by
clearing the destination register and jumping over the faulting
instruction.

A more generic solution would add a "handler" field to the table entry,
like on x86 and s390.

The same issue in ARM64 is fixed in:
commit 800834285361 ("bpf, arm64: Add BPF exception tables")

+#ifdef CONFIG_BPF_JIT
+int rv_bpf_fixup_exception(const struct exception_table_entry *ex, struct pt_regs *regs);
+#endif
+
 int fixup_exception(struct pt_regs *regs)
 {
 	const struct exception_table_entry *fixup;
 
 	fixup = search_exception_tables(regs->epc);
-	if (fixup) {
-		regs->epc = fixup->fixup;
-		return 1;
-	}
-	return 0;
+	if (!fixup)
+		return 0;
+
+#ifdef CONFIG_BPF_JIT
+	if (regs->epc >= BPF_JIT_REGION_START && regs->epc < BPF_JIT_REGION_END)
+		return rv_bpf_fixup_exception(fixup, regs);
+#endif
+
+	regs->epc = fixup->fixup;
+	return 1;
 }

As a heads-up, on the extable front, both arm64 and x86 are moving to
having an enumerated "type" field to select the handler:

x86:

  https://lore.kernel.org/lkml/20210908132525.211958725@linutronix.de/ (local)

arm64:

  https://lore.kernel.org/linux-arm-kernel/20211019160219.5202-11-mark.rutland@arm.com/ (local)

... and going forwards, riscv might want to do likewise.

Thanks,
Mark.