| 12 Apr 2025 |
|  mjolnirchanged room power levels. | 06:52:08 |
| 13 Apr 2025 |
 Andrew | I think it helped, but in the end it now again starts to choke. | 22:49:45 |
| Andrew | I've fixed the graph, the legend was not accurate and the colors not very nice. | 22:50:30 |
| Andrew | Redacted or Malformed Event | 22:50:52 |
| Andrew | ok, I guess I have to wait... | 22:51:26 |
| Andrew | Anyway, here are the new last results after 8.5 days:
81Mi 83424 fs/btrfs/inode.c:7694 [btrfs] func:btrfs_alloc_inode
103Mi 26343 arch/x86/mm/pgtable.c:44 func:pte_alloc_one
219Mi 55821 fs/btrfs/compression.c:477 [btrfs] func:add_ra_bio_pages
220Mi 56232 mm/memory.c:1062 func:folio_prealloc
586Mi 149825 fs/btrfs/extent_io.c:631 [btrfs] func:btrfs_alloc_page_array
598Mi 50415 mm/slub.c:2423 func:alloc_slab_page
848Mi 217053 mm/memory.c:1060 func:folio_prealloc
1.1Gi 280842 mm/readahead.c:186 func:ractl_alloc_folio
2.3Gi 579555 mm/swap_state.c:476 func:__read_swap_cache_async
7.4Gi 8812 drivers/gpu/drm/ttm/ttm_pool.c:97 [ttm] func:ttm_pool_alloc_page
| 22:52:22 |
| Andrew | The umr shows "kernel 7 GB" and ".gnome-shell-wr 6 GB". | 22:53:33 |
| Andrew | * The umr shows "kernel 7 GB" and ".gnome-shell-wr 6 GB". via btop it uses 840 MiB. | 22:54:00 |
| Andrew | Download test.zip | 22:55:29 |
| Andrew | oh, so I can post archives at least. | 22:55:46 |
| Andrew | Download images.zip | 22:57:19 |
| Andrew | There is a drop after 4 days, cuz GNOME started ignoring LMB and I had to log out. | 23:00:18 |
| Andrew | if this is all due to one of the extensions left on, then it's just ridiculous. | 23:05:15 |
| Andrew | I left appindicator on, if it still goes bad, I'm gonna lose it. | 23:13:38 |
| 14 Apr 2025 |
 K900 | Well yeah that's definitely a VRAM leak somewhere | 03:03:41 |
| 19 Apr 2025 |
 FreeVariable | The Gnome 48 PR looks ripe like a fresh fruit! Cannot wait | 17:32:47 |
| Andrew | Is big used + swap says the same thing? Because without cache it's already at 13.6 GiB, with cache it's already over 15 GiB.
total used free shared buff/cache available
Mem: 15Gi 6.3Gi 3.8Gi 107Mi 5.3Gi 8.7Gi
Swap: 15Gi 7.3Gi 8.7Gi
| 21:05:09 |
| Andrew | Download images.zip | 21:05:22 |
| Andrew | 94Mi 4218 mm/slub.c:2425 func:alloc_slab_page
148Mi 37731 mm/memory.c:1062 func:folio_prealloc
304Mi 313650 fs/btrfs/inode.c:7694 [btrfs] func:btrfs_alloc_inode
589Mi 150591 fs/btrfs/compression.c:477 [btrfs] func:add_ra_bio_pages
865Mi 69330 mm/slub.c:2423 func:alloc_slab_page
1.1Gi 269138 fs/btrfs/extent_io.c:631 [btrfs] func:btrfs_alloc_page_array
1.5Gi 1031 drivers/gpu/drm/ttm/ttm_pool.c:97 [ttm] func:ttm_pool_alloc_page
1.6Gi 418610 mm/memory.c:1060 func:folio_prealloc
1.8Gi 446531 mm/swap_state.c:476 func:__read_swap_cache_async
2.9Gi 741395 mm/readahead.c:186 func:ractl_alloc_folio
| 21:05:51 |
| Andrew | It's most definitely now much much better according to all stats, but I worry that if used + swap reach 15 GiB, it would mean that without swap my system will still die. | 21:07:16 |
| Andrew | From the graph and umr for gnome-shell, it looks like there are no more leaks, so, hurray, I think? | 21:07:56 |
|  Ochoa left the room. | 21:30:30 |
| 20 Apr 2025 |
|  @9lore:tchncs.de joined the room. | 03:16:27 |
| K900 | Cache will be released as needed | 06:15:46 |
| K900 | That's why it's cache | 06:15:49 |
| K900 | So was it one of your extensions that was causing the leak? | 06:16:03 |
| Andrew | But what about swap? Surely with a few more days it will increase. Now it's at 7.35 GiB. Or is it if I remove swap, the "used" will not exceed? Like, how does it treat swap? As a spare space to dump more stuff, or does it use absolute minimum to function? | 15:29:08 |
| Andrew | Well, I removed 4 more extensions https://matrix.to/#/!XQQVyIbcAcHFvzmcTl:nixos.org/$zFjjK3ZC8jRObtXlp2gHKZrlEg9kQCqodiy8d24uNU4?via=nixos.org&via=matrix.org&via=tchncs.de | 15:29:17 |
| Andrew | * Well, I removed 4 more extensions https://matrix.to/#/!XQQVyIbcAcHFvzmcTl:nixos.org/$zFjjK3ZC8jRObtXlp2gHKZrlEg9kQCqodiy8d24uNU4?via=nixos.org&via=matrix.org&via=tchncs.de, so I guess some of them were bad. | 15:29:46 |
| Andrew | I really hope it's not the input method one, since it's the most practical and why I now suffer switching between languages. | 15:31:06 |
