pstream: Support memfd blocks transport

Now that we have the necessary infrastructure to memexport and
mempimport a memfd memblock, extend that support higher up in the
chain with pstreams.

A PA endpoint can now _transparently_ send a memfd memblock to the
other end by simply calling pa_pstream_send_memblock() – provided
the block's memfd pool was earlier registered with the pstream.

If the pipe does not support memfd transfers, we fall back to
sending the block's full data instead of just its reference.

** Further details:

A single pstream connection usually transfers blocks from multiple
pools including the server's srbchannel mempool, the client's
audio data mempool, and the server's global core mempool.

If these mempools are memfd-backed, we now require registering
them with the pstream before sending any blocks they cover. This
is done to minimize fd passing overhead and avoid fd leaks.

Moreover, to support all these pools without hard-coding their
number or nature in the Pulse communication protocol itself, a new
REGISTER_MEMFD_SHMID command is introduced. That command can be
sent _anytime_ during the pstream's lifetime and is used for
creating on demand SHM ID to memfd mappings.

Suggested-by: David Henningsson <david.henningsson@canonical.com>
Signed-off-by: Ahmed S. Darwish <darwish.07@gmail.com>

src/pulsecore/pstream-util.c

@@ -21,13 +21,16 @@
 #include <config.h>
 #endif
 
-#include <pulsecore/native-common.h>
+#include <pulsecore/core-util.h>
 #include <pulsecore/macro.h>
+#include <pulsecore/native-common.h>
+#include <pulsecore/pstream.h>
+#include <pulsecore/refcnt.h>
 #include <pulse/xmalloc.h>
 
 #include "pstream-util.h"
 
-static void pa_pstream_send_tagstruct_with_ancil_data(pa_pstream *p, pa_tagstruct *t, const pa_cmsg_ancil_data *ancil_data) {
+static void pa_pstream_send_tagstruct_with_ancil_data(pa_pstream *p, pa_tagstruct *t, pa_cmsg_ancil_data *ancil_data) {
     size_t length;
     const uint8_t *data;
     pa_packet *packet;
@@ -58,12 +61,21 @@ void pa_pstream_send_tagstruct_with_creds(pa_pstream *p, pa_tagstruct *t, const
         pa_pstream_send_tagstruct_with_ancil_data(p, t, NULL);
 }
 
-void pa_pstream_send_tagstruct_with_fds(pa_pstream *p, pa_tagstruct *t, int nfd, const int *fds) {
+/* @close_fds: If set then the pstreams code, after invoking a sendmsg(),
+ * will close all passed fds.
+ *
+ * Such fds cannot be closed here as this might lead to freeing them
+ * before they're actually passed to the other end. The internally-used
+ * pa_pstream_send_packet() does not do any actual writes and just
+ * defers write events over the pstream. */
+void pa_pstream_send_tagstruct_with_fds(pa_pstream *p, pa_tagstruct *t, int nfd, const int *fds,
+                                        bool close_fds) {
     if (nfd > 0) {
         pa_cmsg_ancil_data a;
 
         a.nfd = nfd;
         a.creds_valid = false;
+        a.close_fds_on_cleanup = close_fds;
         pa_assert(nfd <= MAX_ANCIL_DATA_FDS);
         memcpy(a.fds, fds, sizeof(int) * nfd);
         pa_pstream_send_tagstruct_with_ancil_data(p, t, &a);
@@ -78,7 +90,8 @@ void pa_pstream_send_tagstruct_with_creds(pa_pstream *p, pa_tagstruct *t, const
     pa_pstream_send_tagstruct_with_ancil_data(p, t, NULL);
 }
 
-void pa_pstream_send_tagstruct_with_fds(pa_pstream *p, pa_tagstruct *t, int nfd, const int *fds) {
+void pa_pstream_send_tagstruct_with_fds(pa_pstream *p, pa_tagstruct *t, int nfd, const int *fds,
+                                        bool close_fds) {
     pa_assert_not_reached();
 }
 
@@ -102,3 +115,82 @@ void pa_pstream_send_simple_ack(pa_pstream *p, uint32_t tag) {
     pa_tagstruct_putu32(t, tag);
     pa_pstream_send_tagstruct(p, t);
 }
+
+/* Before sending blocks from a memfd-backed pool over the pipe, we
+ * must call this method first.
+ *
+ * This is needed to transfer memfd blocks without passing their fd
+ * every time, thus minimizing overhead and avoiding fd leaks.
+ *
+ * On registration a packet is sent with the memfd fd as ancil data;
+ * such packet has an ID that uniquely identifies the pool's memfd
+ * region. Upon arrival the other end creates a permanent mapping
+ * between that ID and the passed memfd memory area.
+ *
+ * By doing so, we won't need to reference the pool's memfd fd any
+ * further - just its ID. Both endpoints can then close their fds. */
+int pa_pstream_register_memfd_mempool(pa_pstream *p, pa_mempool *pool, const char **fail_reason) {
+#if defined(HAVE_CREDS) && defined(HAVE_MEMFD)
+    unsigned shm_id;
+    int memfd_fd, ret = -1;
+    pa_tagstruct *t;
+    bool per_client_mempool;
+
+    pa_assert(p);
+    pa_assert(fail_reason);
+
+    *fail_reason = NULL;
+    per_client_mempool = pa_mempool_is_per_client(pool);
+
+    pa_pstream_ref(p);
+
+    if (!pa_mempool_is_shared(pool)) {
+        *fail_reason = "mempool is not shared";
+        goto finish;
+    }
+
+    if (!pa_mempool_is_memfd_backed(pool)) {
+        *fail_reason = "mempool is not memfd-backed";
+        goto finish;
+    }
+
+    if (pa_mempool_get_shm_id(pool, &shm_id)) {
+        *fail_reason = "could not extract pool SHM ID";
+        goto finish;
+    }
+
+    if (!pa_pstream_get_memfd(p)) {
+        *fail_reason = "pipe does not support memfd transport";
+        goto finish;
+    }
+
+    memfd_fd = (per_client_mempool) ? pa_mempool_take_memfd_fd(pool) :
+                                      pa_mempool_get_memfd_fd(pool);
+
+    /* Note! For per-client mempools we've taken ownership of the memfd
+     * fd, and we're thus the sole code path responsible for closing it.
+     * In case of any failure, it MUST be closed. */
+
+    if (pa_pstream_attach_memfd_shmid(p, shm_id, memfd_fd)) {
+        *fail_reason = "could not attach memfd SHM ID to pipe";
+
+        if (per_client_mempool)
+            pa_assert_se(pa_close(memfd_fd) == 0);
+        goto finish;
+    }
+
+    t = pa_tagstruct_new();
+    pa_tagstruct_putu32(t, PA_COMMAND_REGISTER_MEMFD_SHMID);
+    pa_tagstruct_putu32(t, (uint32_t) -1); /* tag */
+    pa_tagstruct_putu32(t, shm_id);
+    pa_pstream_send_tagstruct_with_fds(p, t, 1, &memfd_fd, per_client_mempool);
+
+    ret = 0;
+finish:
+    pa_pstream_unref(p);
+    return ret;
+
+#else
+    pa_assert_not_reached();
+#endif
+}