summaryrefslogtreecommitdiff
path: root/arch
diff options
context:
space:
mode:
authorLinus Torvalds <torvalds@linux-foundation.org>2020-08-04 14:40:07 -0700
committerLinus Torvalds <torvalds@linux-foundation.org>2020-08-04 14:40:07 -0700
commit0a72761b27fe3b10e3f336bf2f2aa22635504cdd (patch)
treefbd99b0eac76464388070ef29153b9c8d596dc5f /arch
parent3950e975431bc914f7e81b8f2a2dbdf2064acb0f (diff)
parent55d9ad97e417cc2604654913e902d26f942bde00 (diff)
Merge tag 'threads-v5.9' of git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux
Pull thread updates from Christian Brauner: "This contains the changes to add the missing support for attaching to time namespaces via pidfds. Last cycle setns() was changed to support attaching to multiple namespaces atomically. This requires all namespaces to have a point of no return where they can't fail anymore. Specifically, <namespace-type>_install() is allowed to perform permission checks and install the namespace into the new struct nsset that it has been given but it is not allowed to make visible changes to the affected task. Once <namespace-type>_install() returns, anything that the given namespace type additionally requires to be setup needs to ideally be done in a function that can't fail or if it fails the failure must be non-fatal. For time namespaces the relevant functions that fell into this category were timens_set_vvar_page() and vdso_join_timens(). The latter could still fail although it didn't need to. This function is only implemented for vdso_join_timens() in current mainline. As discussed on-list (cf. [1]), in order to make setns() support time namespaces when attaching to multiple namespaces at once properly we changed vdso_join_timens() to always succeed. So vdso_join_timens() replaces the mmap_write_lock_killable() with mmap_read_lock(). Please note that arm is about to grow vdso support for time namespaces (possibly this merge window). We've synced on this change and arm64 also uses mmap_read_lock(), i.e. makes vdso_join_timens() a function that can't fail. Once the changes here and the arm64 changes have landed, vdso_join_timens() should be turned into a void function so it's obvious to callers and implementers on other architectures that the expectation is that it can't fail. We didn't do this right away because it would've introduced unnecessary merge conflicts between the two trees for no major gain. As always, tests included" [1]: https://lore.kernel.org/lkml/20200611110221.pgd3r5qkjrjmfqa2@wittgenstein * tag 'threads-v5.9' of git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux: tests: add CLONE_NEWTIME setns tests nsproxy: support CLONE_NEWTIME with setns() timens: add timens_commit() helper timens: make vdso_join_timens() always succeed
Diffstat (limited to 'arch')
-rw-r--r--arch/x86/entry/vdso/vma.c5
1 files changed, 2 insertions, 3 deletions
diff --git a/arch/x86/entry/vdso/vma.c b/arch/x86/entry/vdso/vma.c
index ea7c1f0b79df..9185cb1d13b9 100644
--- a/arch/x86/entry/vdso/vma.c
+++ b/arch/x86/entry/vdso/vma.c
@@ -144,8 +144,7 @@ int vdso_join_timens(struct task_struct *task, struct time_namespace *ns)
struct mm_struct *mm = task->mm;
struct vm_area_struct *vma;
- if (mmap_write_lock_killable(mm))
- return -EINTR;
+ mmap_read_lock(mm);
for (vma = mm->mmap; vma; vma = vma->vm_next) {
unsigned long size = vma->vm_end - vma->vm_start;
@@ -154,7 +153,7 @@ int vdso_join_timens(struct task_struct *task, struct time_namespace *ns)
zap_page_range(vma, vma->vm_start, size);
}
- mmap_write_unlock(mm);
+ mmap_read_unlock(mm);
return 0;
}
#else