On Mon, 2026-09-28 at 04:47 -0700, Dora, Sunil Kumar via lists.openembedded.org 
wrote:
> I might be missing something, but I think the aarch64 case can still leak. 
> In configure.ac, right after the cached sysconf check there's:
> 
>   AS_CASE([$build_cpu], [aarch64*], ... erofs_cv_max_block_size=16384 ...)
> 
> so on an aarch64 worker building qemux86-64 the cached 4096 still gets bumped 
> to 16384. 
> When I tried the v1.9.4 configure with --build=aarch64-linux 
> --host=x86_64-linux, config.h ended up with 16384.
> 
> Passing MAX_BLOCK_SIZE=${BLOCKSIZE} instead seems to skip that whole block. 
> The next 119 run should show it, as long as reproducibleA picks up 
> erofs-utils from x86_64 sstate.

You may well be correct. I'd isolated it to the blocksize and the patch
is in testing but might not fully work :(. I was planning to verify
before merging. I sent it as it at least points at where the issue
is...

It sounds like setting MAX_BLOCK_SIZE instead would be the thing to do
as you say.

Cheers,

Richard

-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#246765): 
https://lists.openembedded.org/g/openembedded-core/message/246765
Mute This Topic: https://lists.openembedded.org/mt/121458039/21656
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub 
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to