github-mirror/linux
1
0
Fork 0
mirror of https://github.com/torvalds/linux.git synced 2026-11-07 06:17:02 +01:00
Code Issues Actions Packages Projects Releases Wiki Activity Graph

bpf: Allow ksock connect/send calls from wq

Browse Source 
The initial ksock implementation, commit 7ae4eb14c5 ("bpf: Add ksock
kfuncs"), kept LSM checks from the kfunc calls and disallowed
asynchronous workqueues because worker credentials differ from those of
the triggering task. However, discussions at Plumbers 2026[^1] revealed
that bypassing LSM checks for BPF-created sockets could be acceptable.

This patch allows workqueue use while retaining LSM checks for now,
paving the way to call bpf_ksock_connect() and bpf_ksock_send() from
non-sleepable programs, such as tracing programs, in future patches.

Note that bpf_ksock_create() still checks:
* for current to not be a kernel thread (thus excluding bpf workqueue
  callbacks) because the socket is created within a userspace task
  network namespace. It should, most of the time, be the process that
  triggered the SYSCALL program, but could be another task when called
  from bpf_task_work callbacks.
* for current->nsproxy to be non-NULL since when called from a
  bpf_task_work callback and the task is in the do_exit() phase, nsproxy
  is torn down just before the remaining task work is executed.

[^1]: https://lpc.events/event/20/contributions/2430/

Signed-off-by: Mahe Tardy <mahe.tardy@gmail.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Tested-by: Alan Maguire (Oracle) <almagui@kernel.org>
Link: https://patch.msgid.link/20261016133318.33999-1-mahe.tardy@gmail.com
This commit is contained in:
Mahe Tardy 2026-10-16 13:33:17 +00:00 committed by Alexei Starovoitov
parent aaae5e500a
commit 60bbe76a3f
No known key found for this signature in database
1 changed files with 4 additions and 20 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
net/core/bpf_ksock.c
View File

@ -37,15 +37,6 @@ static void ksock_release_work_fn(struct work_struct *work)
kfree(ks);
}
static bool bpf_ksock_has_user_task_context(void)
{
/*
* Task work can run from do_exit() after exit_nsproxy_namespaces()
* cleared current->nsproxy, while current is still not a kthread.
*/
return !(current->flags & PF_KTHREAD) && current->nsproxy;
}
__bpf_kfunc_start_defs();
/**
@ -59,7 +50,7 @@ __bpf_kfunc_start_defs();
* This function may sleep (sock_create), so it can only be used
* in sleepable BPF programs (SYSCALL).
* It cannot be called from a BPF workqueue callback because that callback
* does not retain the invoking task's namespace or security context.
* does not retain the invoking task's namespace.
*
* @opts: Pointer to struct bpf_ksock_create_opts with socket parameters.
* @opts__sz: Size of the opts struct.
@ -74,11 +65,10 @@ bpf_ksock_create(const struct bpf_ksock_create_opts *opts, u32 opts__sz,
int err;
/*
* sock_create() derives the network namespace, credentials, and cgroup
* from current. Kernel threads, including BPF workqueue callbacks, do
* not carry the context of the task that invoked the BPF program.
* sock_create() derives the network namespace from current and task
* work can run from do_exit() after exit_nsproxy_namespaces()
*/
if (!bpf_ksock_has_user_task_context()) {
if ((current->flags & PF_KTHREAD) || !current->nsproxy) {
err = -EOPNOTSUPP;
goto err_out;
}
@ -165,9 +155,6 @@ __bpf_kfunc int bpf_ksock_connect(struct bpf_ksock *ks,
struct sockaddr_storage sa;
int addrlen;
if (!bpf_ksock_has_user_task_context())
return -EOPNOTSUPP;
if (!addr || addr__sz != sizeof(*addr))
return -EINVAL;
@ -247,9 +234,6 @@ __bpf_kfunc int bpf_ksock_send(struct bpf_ksock *ks, const void *data,
};
int ret;
if (!bpf_ksock_has_user_task_context())
return -EOPNOTSUPP;
ret = kernel_sendmsg(ks->sock, &msg, &iov, 1, data__sz);
return ret;