On Monday 09 December 2024 16:28:08 Paulo Alcantara wrote: > Pali Rohár <pali@xxxxxxxxxx> writes: > > > Hello, this patch is incomplete and still does not fix the main problem > > that SMB2_OP_QUERY_WSL_EA command does not work with any Windows SMB > > server except the last Windows Server version. On non-recent Windows > > versions it is not possible to set both EAs and reparse point at the > > same time and Windows SMB server is returning error when trying to query > > EAs on file with reparse point. > > No, Ralph's patch has nothing to do with your problem. SMB3.1.1 POSIX > will support NFS reparse points with no EAs, so makes sense skipping > SMB2_OP_QUERY_WSL_EA altogether for posix case. Ok, so this is just a coincidence that skip was added at the same place. And I though that it is the same thing (which now I see that is not). > > Which basically means that it is not possible to query data about the > > special files from Windows SMB server (except 2022 version). > > > > More details are in the email which I wrote in September: > > https://lore.kernel.org/linux-cifs/20240928140939.vjndryndfngzq7x4@pali/ > > > > I proposed similar but extended patch which skips asking for EAs based > > on reparse point: > > https://lore.kernel.org/linux-cifs/20240913200204.10660-1-pali@xxxxxxxxxx/ > > Yes, that patch looks incorrect and untested. Can you tell me how is > @data->reparse.tag supposed to be set, for non-readdir case, if the > compound request wasn't sent yet? Have you tried to stat(2) those files > with your patch? I need to rethink about it. Basically I already dropped this patch as I was expecting the proper solution. > > But it was somehow rejected as the proper solution should be different: > > https://lore.kernel.org/linux-cifs/20240917210707.4lt4obty7wlmm42j@pali/ > > Yes. Have you sent a patch with the proposed solution yet? I have not sent, I was not working on it. I was in impression that you are going to implement the proper solution.
