On 11/8/23 5:20 PM, Eduard Zingerman wrote:
On Wed, 2023-11-08 at 12:05 -0800, Alexei Starovoitov wrote:
[...]
The algorithm doesn't look simple.
Even if we change llvm to do this, it's not clear whether
the verifier will be able to consume such code.
Actually, I don't think that trampoline jumps could cause any trouble.
imo it's too much effort to address a non-issue.
I'd just adjust the pyperf180.c test.
Ok, I'll drop this. Thank you for taking a look.
Thanks Eduard for doing analysis for this! I will send
a patch soon to fix selftest failure issue.
