On Tue, 21 Jan 2025 at 14:53, James Bottomley <James.Bottomley@xxxxxxxxxxxxxxxxxxxxx> wrote: > > On Mon, 2025-01-20 at 13:57 -0500, James Bottomley wrote: > > On Mon, 2025-01-20 at 17:31 +0100, Ard Biesheuvel wrote: > > > On Sun, 19 Jan 2025 at 17:59, Ard Biesheuvel <ardb@xxxxxxxxxx> > > > wrote: > > > > > > > > On Sun, 19 Jan 2025 at 16:12, James Bottomley > > > > <James.Bottomley@xxxxxxxxxxxxxxxxxxxxx> wrote: > > > > > > > > ... > > > > > > > > Thanks James. I've queued up this version now, so we'll get some > > > > coverage from the robots. I'll redo my own testing tomorrow, but > > > > I'll omit these changes from my initial PR to Linus. If we're > > > > confident that things are sound, I'll send another PR during the > > > > second half of the merge window. > > > > > > I'm hitting the failure cases below. The first one appears to hit > > > the same 'Operation not permitted' condition on the write, the > > > error message is just hidden by the /dev/null redirect. > > > > > > I'm running the make command from a root shell. Using printf from > > > the command line works happily so I suspect there is some issue > > > with the concurrency and the subshells? > > > > It could be that the file isn't opened until the subshell is spawned. > > I can probably use the pipe to wait for the subshell to start; I'll > > try to code that up. > > OK, this is what I came up with. It works for me (but then so did the > other script ... I think my testing VM is probably a little slow). > Thanks - that works on my end too. I am testing this on a bare metal arm64 machine, but an ancient and very slow one. So it might be the other way around. I am going to grab this version and queue it up. Thanks.
