On Wed, 27 Oct 2021 at 13:03, Tong Tiangen [off-list ref] 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")

Signed-off-by: Tong Tiangen <redacted>
Tested-by: Pu Lehui <pulehui@huawei.com>
---
v3:
Modify according to Björn's comments, mainly code optimization.

Thank you!

I ran this patch against the test_bpf.ko, and selftests/bpf -- no
regressions, and after the patch is applied more tests passes. Yay!

On a related note. The RISC-V selftests/bpf is in a pretty lousy
state. I'll send a cleanup patch for them soonish. E.g.:

* RISC-V is missing in bpf_tracing.h (libbpf)
* Some programs don't converge in 16 steps, I had to increase it to ~32
* The selftest/bpf Makefile needed some RV specific changes
* ...a lot of tests still don't pass, and needs to be looked in to

Feel free to add:

Acked-by: Björn Töpel <bjorn@kernel.org>

v2:
Modify according to Björn's comments, mainly removes redundant head files
extable.h and some code style issues.

 arch/riscv/mm/extable.c         |  19 +++-
 arch/riscv/net/bpf_jit.h        |   1 +
 arch/riscv/net/bpf_jit_comp64.c | 185 +++++++++++++++++++++++++-------
 arch/riscv/net/bpf_jit_core.c   |  19 ++--
 4 files changed, 177 insertions(+), 47 deletions(-)

[...]

Björn