On 5/14/2026 11:35 AM, Ard Biesheuvel via groups.io wrote:
On Thu, 14 May 2026, at 20:11, Oliver Smith-Denny wrote:
On 5/13/2026 8:50 AM, Oliver Smith-Denny wrote:
On 5/13/2026 3:42 AM, Ard Biesheuvel via groups.io wrote:
On Tue, 12 May 2026, at 22:02, Oliver Smith-Denny via groups.io wrote:
(Untested attempt below)
...
Doug and I started down this path yesterday as well, I'll do some
testing of this today (and compare our version to yours, make sure we
didn't miss something).
This worked in our testing, both in the original crypto API failure
case and a pair of test drivers I wrote that communicate via protocol,
calling variadic functions (which should always work, but I wanted to
make sure) and passing VA_LIST as a parameter. Each side validated the
expected input/output. I tried all four combinations of GCC vs CLANGPDB
built consumer/producer and all worked (without this change all
four combinations were broken).
VA_LIST or VA_LIST* ?
I had originally tested with VA_LIST because that was the failing case,
but because VA_LIST* would be the correct version I also just tested
that and it works in all cases.
Doug is going to put up a change for VA_LIST -> VA_LIST * separately.
Do you want to put up a PR for this? I can, but I didn't make any
changes to your suggestion, it "just worked." Feel free to add a
tested-by for me, not that carries too much weight :)
Excellent!
Feel free to propose a pr, and credit/blame me in the commit log 😃
Great, I'll put that up now!
Oliver
-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#121954): https://edk2.groups.io/g/devel/message/121954
Mute This Topic: https://groups.io/mt/119278849/21656
Group Owner: [email protected]
Unsubscribe: https://edk2.groups.io/g/devel/unsub [[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-
