diff options
author | Christian Brauner <brauner@kernel.org> | 2025-09-15 13:45:37 +0200 |
---|---|---|
committer | Christian Brauner <brauner@kernel.org> | 2025-09-19 14:26:17 +0200 |
commit | 3ab378cfa793c648d4edf02bbfff3af8715aca91 (patch) | |
tree | b5a57da7ff6c26434569ec95e8d2cf43f9cbe3c6 /rust/kernel/alloc/allocator.rs | |
parent | 8f5ae30d69d7543eee0d70083daf4de8fe15d585 (diff) | |
parent | 28ef38a9a2c7aa4dd8dfbb1d2b12f9ea84044327 (diff) |
Merge patch series "ns: support file handles"
Christian Brauner <brauner@kernel.org> says:
For a while now we have supported file handles for pidfds. This has
proven to be very useful.
Extend the concept to cover namespaces as well. After this patchset it
is possible to encode and decode namespace file handles using the
commong name_to_handle_at() and open_by_handle_at() apis.
Namespaces file descriptors can already be derived from pidfds which
means they aren't subject to overmount protection bugs. IOW, it's
irrelevant if the caller would not have access to an appropriate
/proc/<pid>/ns/ directory as they could always just derive the namespace
based on a pidfd already.
It has the same advantage as pidfds. It's possible to reliably and for
the lifetime of the system refer to a namespace without pinning any
resources and to compare them.
Permission checking is kept simple. If the caller is located in the
namespace the file handle refers to they are able to open it otherwise
they must hold privilege over the owning namespace of the relevant
namespace.
Both the network namespace and the mount namespace already have an
associated cookie that isn't recycled and is fully exposed to userspace.
Move this into ns_common and use the same id space for all namespaces so
they can trivially and reliably be compared.
There's more coming based on the iterator infrastructure but the series
is large enough and focuses on file handles.
Extensive selftests included.
* patches from https://lore.kernel.org/20250912-work-namespace-v2-0-1a247645cef5@kernel.org: (33 commits)
selftests/namespaces: add file handle selftests
selftests/namespaces: add identifier selftests
tools: update nsfs.h uapi header
nsfs: add missing id retrieval support
nsfs: support exhaustive file handles
nsfs: support file handles
nsfs: add current_in_namespace()
ns: add to_<type>_ns() to respective headers
uts: support ns lookup
user: support ns lookup
time: support ns lookup
pid: support ns lookup
net: support ns lookup
ipc: support ns lookup
cgroup: support ns lookup
mnt: support ns lookup
nstree: make iterator generic
ns: remove ns_alloc_inum()
uts: use ns_common_init()
user: use ns_common_init()
...
Signed-off-by: Christian Brauner <brauner@kernel.org>
Diffstat (limited to 'rust/kernel/alloc/allocator.rs')
-rw-r--r-- | rust/kernel/alloc/allocator.rs | 30 |
1 files changed, 18 insertions, 12 deletions
diff --git a/rust/kernel/alloc/allocator.rs b/rust/kernel/alloc/allocator.rs index aa2dfa9dca4c..2692cf90c948 100644 --- a/rust/kernel/alloc/allocator.rs +++ b/rust/kernel/alloc/allocator.rs @@ -43,17 +43,6 @@ pub struct Vmalloc; /// For more details see [self]. pub struct KVmalloc; -/// Returns a proper size to alloc a new object aligned to `new_layout`'s alignment. -fn aligned_size(new_layout: Layout) -> usize { - // Customized layouts from `Layout::from_size_align()` can have size < align, so pad first. - let layout = new_layout.pad_to_align(); - - // Note that `layout.size()` (after padding) is guaranteed to be a multiple of `layout.align()` - // which together with the slab guarantees means the `krealloc` will return a properly aligned - // object (see comments in `kmalloc()` for more information). - layout.size() -} - /// # Invariants /// /// One of the following: `krealloc`, `vrealloc`, `kvrealloc`. @@ -88,7 +77,7 @@ impl ReallocFunc { old_layout: Layout, flags: Flags, ) -> Result<NonNull<[u8]>, AllocError> { - let size = aligned_size(layout); + let size = layout.size(); let ptr = match ptr { Some(ptr) => { if old_layout.size() == 0 { @@ -123,6 +112,17 @@ impl ReallocFunc { } } +impl Kmalloc { + /// Returns a [`Layout`] that makes [`Kmalloc`] fulfill the requested size and alignment of + /// `layout`. + pub fn aligned_layout(layout: Layout) -> Layout { + // Note that `layout.size()` (after padding) is guaranteed to be a multiple of + // `layout.align()` which together with the slab guarantees means that `Kmalloc` will return + // a properly aligned object (see comments in `kmalloc()` for more information). + layout.pad_to_align() + } +} + // SAFETY: `realloc` delegates to `ReallocFunc::call`, which guarantees that // - memory remains valid until it is explicitly freed, // - passing a pointer to a valid memory allocation is OK, @@ -135,6 +135,8 @@ unsafe impl Allocator for Kmalloc { old_layout: Layout, flags: Flags, ) -> Result<NonNull<[u8]>, AllocError> { + let layout = Kmalloc::aligned_layout(layout); + // SAFETY: `ReallocFunc::call` has the same safety requirements as `Allocator::realloc`. unsafe { ReallocFunc::KREALLOC.call(ptr, layout, old_layout, flags) } } @@ -176,6 +178,10 @@ unsafe impl Allocator for KVmalloc { old_layout: Layout, flags: Flags, ) -> Result<NonNull<[u8]>, AllocError> { + // `KVmalloc` may use the `Kmalloc` backend, hence we have to enforce a `Kmalloc` + // compatible layout. + let layout = Kmalloc::aligned_layout(layout); + // TODO: Support alignments larger than PAGE_SIZE. if layout.align() > bindings::PAGE_SIZE { pr_warn!("KVmalloc does not support alignments larger than PAGE_SIZE yet.\n"); |