In the Linux kernel, the following vulnerability has been resolved:<\/p>\n
memcg: avoid charging the root memcg from obj_cgroup_charge_pages()<\/p>\n
obj_cgroup_charge_pages() resolves the objcg to its memcg and calls
\ntry_charge_memcg(), which does not short circuit the root memcg. That
\nmemcg can be the root memcg: obj_cgroup_is_root() reflects the memcg the
\nobjcg was created for and is never updated, while memcg_reparent_objcgs()
\ndoes redirect objcg->memcg to the parent on rmdir. An objcg of a dying
\nchild of root therefore passes every obj_cgroup_is_root() filter but
\nresolves to the root memcg.<\/p>\n
Folios keep the objcg they were charged with, so this is easy to reach
\nthrough zswap: allocate anon memory in a cgroup, move the task out, remove
\nthe cgroup, then write to the root cgroup’s memory.reclaim. The reclaimed
\nfolios are charged through the reparented objcg and end up in
\nrefill_stock() with the root memcg:<\/p>\n
WARNING: mm\/memcontrol.c:2198 at refill_stock+0x644\/0x940
\n refill_stock+0x644\/0x940
\n try_charge_memcg+0x12d6\/0x1570
\n __obj_cgroup_charge+0x35\/0xf0
\n obj_cgroup_charge+0x1de\/0x210
\n obj_cgroup_charge_zswap+0x83\/0x270
\n zswap_store+0x1620\/0x2000
\n swap_writeout+0x94c\/0x14c0
\n shrink_folio_list+0x3388\/0x52b0
\n […]
\n try_to_free_mem_cgroup_pages+0x30d\/0x830
\n user_proactive_reclaim+0x504\/0x840
\n memory_reclaim+0x1f\/0x30<\/p>\n
Beyond the warning, the charge is asymmetric: obj_cgroup_uncharge_pages()
\nskips refill_stock() for the root memcg, so the root’s page counter grows
\nand is never uncharged. It is not user visible, since memory.current is
\nnot exposed on the root, but it is a leak.<\/p>\n
Use try_charge(), which returns early for the root memcg, restoring the
\nsymmetry with obj_cgroup_uncharge_pages().<\/p>\n
The above sequence was scripted into a standalone reproducer (zswap on,
\nswap on a virtio disk, 512MB of anon memory faulted in inside a child of
\nthe root cgroup, the task then migrated to the root cgroup, the child
\nremoved, followed by “echo 600M swappiness=max > memory.reclaim” on the
\nroot) and run in a CONFIG_DEBUG_VM=y VM. It reproduces the splat on the
\nfirst zswap store of a reparented folio, with the same call chain as the
\nreport. With this patch applied the splat is gone while the zswap store
\ncount over the run is unchanged, so the same path is still exercised.
\ncgroup selftests test_zswap, test_kmem and test_memcontrol show no new
\nfailures.<\/p>\n","protected":false},"excerpt":{"rendered":"
In the Linux kernel, the following vulnerability has been resolved:<\/p>\n
memcg: avoid charging the root memcg from obj_cgroup_charge_pages()<\/p>\n
obj_cgroup_charge_pages() resolves the objcg to its memcg and <\/p>\n","protected":false},"template":"","meta":[],"categories":[],"tags":[],"class_list":["post-39085","vuln","type-vuln","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=\/wp\/v2\/vuln\/39085","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=\/wp\/v2\/vuln"}],"about":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=\/wp\/v2\/types\/vuln"}],"wp:attachment":[{"href":"https:\/\/www.54iter.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=39085"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.54iter.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=39085"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.54iter.cn\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=39085"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}