Re: [PATCH] selftests/ftrace: update kprobe syntax error test for ppc64le

[Hari Bathini <hbathini@linux.ibm.com>]

On 02/11/24 2:29 am, Segher Boessenkool wrote:
> Hi!
> 
> On Sat, Nov 02, 2024 at 12:49:25AM +0530, Hari Bathini wrote:
>> For ppc64le, depending on the kernel configuration used, offset 16
>> from function start address can also be considered function entry.
>> Update the test case to accommodate such configurations.
> 
> (This is true for all ELfv2, not just LE.  For the kernel that is about
> the same).
> 
> The LEP and GEP can differ by zero, one, two, four, eight, or sixteen
> insns (where an insn is four bytes).  Four insns is common, yes, but
> maybe you can support all?  See the function symbol's st_other field
> to see what the offset is:
> 0, 1: zero insns, zero bytes
> N = 2..6: 1 << (N-2) insns, i.e. 1<<N bytes
> 7: reserved
> 
> (This is the top 3 bits of st_other, the other bits have other meanings).
> 
> Four insns is common, yes, but by no means the only possibility.

Hi Segher,

Querying for function arguments is supported on kprobes only at function
entry. This is a negative test case where the offset is intentionally
set beyond function entry while querying for function arguments.
I guess, simply setting the offset to 20 (vfs_read is anyway
going to be beyond 5 instructions) instead of 8 for powerpc would
make all platforms and ABI variants happy?

Thanks
Hari