| Moc | Who should I talk to regarding a packaging script error with Nvidia driver forcing a lot of user to use a old driver (340) rather than a newer one like 390 ? | 01:01 |
|---|---|---|
| roaksoax | /win 14 | 08:46 |
| federico3 | hi | 11:47 |
| tdaitx | @cjwatson I'm taking a look at doing an xenial+bionic sru for LP: #1667512, but I'm curious on why doing a "filesystem" sync was preferred compared to a "file" sync (ie. "sync -f <initrd.img>" instead of "sync <initrd.img>") | 13:19 |
| tdaitx | the particular patch you proposed is at https://salsa.debian.org/kernel-team/initramfs-tools/merge_requests/6/diffs | 13:19 |
| udevbot_ | Error: "cjwatson" is not a valid command. | 13:19 |
| tdaitx | cjwatson: ^ | 13:20 |
| cjwatson | tdaitx: I think I simply didn't realise that the file mode existed - the man page doesn't make it clear | 13:35 |
| cjwatson | tdaitx: That said, the fsync mode is tricky - you have to fsync the containing directory as well | 13:36 |
| cjwatson | tdaitx: So I think syncfs is harder to get wrong | 13:36 |
| tdaitx | cjwatson: hmm, you mean "sync the containing directory" because it is a new file, right? in that case, yes, syncfs is better | 14:03 |
| cjwatson | indeed | 14:04 |
| cjwatson | don't want to end up with the dirent still pointing to the previous iteration of the same initramfs or something | 14:04 |
| tdaitx | indeed, fsync for new files is filesystem dependent | 14:06 |
| bneff | Having a weird issue where after using the keyboard audio controls, chrome does not process any mouse clicks. Any pointers? | 17:21 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!