| Summary: | sys-apps/portage-9999 unable to utilise 'portage' user home directory | ||
|---|---|---|---|
| Product: | Portage Development | Reporter: | Joe Kappus <joe> |
| Component: | Core - Interface (emerge) | Assignee: | Portage team <dev-portage> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | james.al.stevenson, mgorny, sam |
| Priority: | Normal | ||
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux | ||
| See Also: | https://bugs.gentoo.org/show_bug.cgi?id=674212 | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Attachments: | build.log | ||
|
Description
Joe Kappus
2021-09-07 06:32:51 UTC
…misspelled? Could you be kind enough to add some print()s to figure out which part doesn't work on your system? probably ugly way to go about it but I added this and then merged a package:
print (os.path.expanduser("~portage"))
print (os.access(os.path.expanduser("~portage"), os.R_OK))
print (os.access(os.path.expanduser("~portage"), os.W_OK))
print (os.access(os.path.expanduser("~portage"), os.X_OK))
Results were:
/var/lib/portage/home
True
False
True
So it can't write to that directory. To verify permissions I then ran these:
stat /var/lib/portage/home/
File: /var/lib/portage/home/
Size: 39 Blocks: 0 IO Block: 4096 directory
Device: 10301h/66305d Inode: 838586180 Links: 2
Access: (0777/drwxrwxrwx) Uid: ( 250/ portage) Gid: ( 250/ portage)
Access: 2020-12-27 21:32:19.462124992 -0500
Modify: 2021-05-11 05:21:51.837840580 -0400
Change: 2021-09-07 02:14:26.829968504 -0400
Birth: 2020-12-27 21:32:19.462124992 -0500
stat /var/lib/portage/
File: /var/lib/portage/
Size: 94 Blocks: 0 IO Block: 4096 directory
Device: 10301h/66305d Inode: 537354957 Links: 3
Access: (2755/drwxr-sr-x) Uid: ( 250/ portage) Gid: ( 250/ portage)
Access: 2018-01-20 16:04:43.000000000 -0500
Modify: 2021-09-07 02:38:32.083304527 -0400
Change: 2021-09-07 02:38:32.083304527 -0400
Birth: 2020-03-06 16:16:58.644445645 -0500
(In reply to Joe Kappus from comment #3) Was the process running as the "portage" user when those tests were performed? emerge was run as root, all the processes it spawned were under portage user. Hmm... maybe interference from sys-apps/sandbox? I don't think so. With FEATURES="-sandbox" it still happens. >stat /var/lib/portage/ > ... >Access: (2755/drwxr-sr-x) Uid: ( 250/ portage) Gid: ( 250/ portage) /var/lib/portage on my system is owned by root:portage, 2755 perms. >Access: 2018-01-20 16:04:43.000000000 -0500 >Modify: 2021-09-07 02:38:32.083304527 -0400 >Change: 2021-09-07 02:38:32.083304527 -0400 > Birth: 2020-03-06 16:16:58.644445645 -0500 Stats seem a little screwy, accessed on 2018 but not birthed until 2020 ??? (In reply to Joe Kappus from comment #7) > I don't think so. With FEATURES="-sandbox" it still happens. Try with FEATURES="-usersandbox". (In reply to Don O from comment #8) > Stats seem a little screwy, accessed on 2018 but not birthed until 2020 ??? Probably the filesystem is mounted with the "noatime" option. Don, I mount with noatime so that's expected to see some weirdness like this. This install was copied from another drive so presumably birth time is when it was copied into the new drive, which happened after the last time I mounted with atimes. Mike, needed both -sandbox and -usersandbox in FEATURES, but it finally merged without warning. So I guess it is sandbox related.... (In reply to Joe Kappus from comment #11) > Mike, needed both -sandbox and -usersandbox in FEATURES, but it finally > merged without warning. So I guess it is sandbox related.... Ok, that makes sense. /var/lib/portage/home is probably not configured as a writable path in the sandbox config, so libsandbox denies any attempt to write there. In any case, the offending commit was reverted earlier today. https://gitweb.gentoo.org/proj/portage.git/commit/?id=3ccfd9a375d34b2c2731064ea4bc9e1a4b8946f3 |
