On 06.08.2025 13:01, Philippe Mathieu-Daudé wrote:
On 6/8/25 09:56, Michael Tokarev wrote:
This is what happens when rushing for last-minute
pre-release fixes.. I'm sorry for this.
My bad actually :/
Well, we both are guilty here :)
And I'm probably more guilty because my initial fix was testing the
same as the actual code is using, and I felt that your suggestion is
wrong - I had everything to at least *check* the damn thing closely.
But I didn't do that.
It is more, - I felt that with your condition, the test might be skipped
even on o_direct-supporting filesystem. But I didn't verify that *too*,
which is even worse.
Instead, I blindly took your advice, without verifying my own doubts.
I think this is more wrong than, when one have serious doubts but does
not bother even verifying them.
Now, back to work..
Philippe, R-b? :)
Thanks,
/mjt
