libbpf: fixes for AF_XDP teardown
Subsystems (4)
cover libbpf: fixes for AF_XDP teardown
William found two bugs, when doing socket teardown within the same process.
The first issue was an invalid munmap call, and the second one was an invalid XSKMAP cleanup. Both resulted in that the process kept references to the socket, which was not correctly cleaned up. When a new socket was created, the bind() call would fail, since the old socket was still lingering, refusing to give up the queue on the netdev.
More details can be found in the individual commits.
Thanks,
Björn
Björn Töpel (2):
libbpf: fix invalid munmap call
libbpf: proper XSKMAP cleanup
tools/lib/bpf/xsk.c | 192 +++++++++++++++++++++++--------------------- 1 file changed, 100 insertions(+), 92 deletions(-)
Signature · 1 line
2.20.1
PATCH 1/2 libbpf: fix invalid munmap call
From: Björn Töpel <[email protected]>
When unmapping the AF_XDP memory regions used for the rings, an invalid address was passed to the munmap() calls. Instead of passing the beginning of the memory region, the descriptor region was passed to munmap.
When the userspace application tried to tear down an AF_XDP socket, the operation failed and the application would still have a reference to socket it wished to get rid of.
--- tools/lib/bpf/xsk.c | 77 +++++++++++++++++++++++---------------------- 1 file changed, 40 insertions(+), 37 deletions(-)
tools/lib/bpf/xsk.c +40 -37
@@ -248,8 +248,7 @@ int xsk_umem__create(struct xsk_umem **umem_ptr, void *umem_area, __u64 size,return 0 ; out_mmap: -munmap (umem->fill, - off.fr.desc + umem->config.fill_size *sizeof (__u64)); +munmap (map, off.fr.desc + umem->config.fill_size *sizeof (__u64)); out_socket:close (umem->fd); out_umem_alloc:
@@ -523,11 +522,11 @@ int xsk_socket__create(struct xsk_socket **xsk_ptr, const char *ifname, struct xsk_ring_cons *rx, struct xsk_ring_prod *tx, const struct xsk_socket_config *usr_config) { + void *rx_map = NULL, *tx_map = NULL;struct sockaddr_xdp sxdp = {};struct xdp_mmap_offsets off;struct xsk_socket *xsk;socklen_t optlen; -void *map;int err;if (!umem || !xsk_ptr || !rx || !tx)
@@ -593,40 +592,40 @@ int xsk_socket__create(struct xsk_socket **xsk_ptr, const char *ifname, }if (rx) { - map =xsk_mmap (NULL , off.rx.desc + - xsk->config.rx_size *sizeof (struct xdp_desc ), -PROT_READ |PROT_WRITE , -MAP_SHARED |MAP_POPULATE , - xsk->fd,XDP_PGOFF_RX_RING ); -if (map ==MAP_FAILED ) { + rx_map =xsk_mmap (NULL , off.rx.desc + + xsk->config.rx_size *sizeof (struct xdp_desc ), +PROT_READ |PROT_WRITE , +MAP_SHARED |MAP_POPULATE , + xsk->fd,XDP_PGOFF_RX_RING ); +if (rx_map ==MAP_FAILED ) { err = -errno; goto out_socket; } rx->mask = xsk->config.rx_size -1 ; rx->size = xsk->config.rx_size; - rx->producer = map + off.rx.producer; - rx->consumer = map + off.rx.consumer; - rx->ring = map + off.rx.desc; + rx->producer = rx_map + off.rx.producer; + rx->consumer = rx_map + off.rx.consumer; + rx->ring = rx_map + off.rx.desc; } xsk->rx = rx;if (tx) { - map =xsk_mmap (NULL , off.tx.desc + - xsk->config.tx_size *sizeof (struct xdp_desc ), -PROT_READ |PROT_WRITE , -MAP_SHARED |MAP_POPULATE , - xsk->fd,XDP_PGOFF_TX_RING ); -if (map ==MAP_FAILED ) { + tx_map =xsk_mmap (NULL , off.tx.desc + + xsk->config.tx_size *sizeof (struct xdp_desc ), +PROT_READ |PROT_WRITE , +MAP_SHARED |MAP_POPULATE , + xsk->fd,XDP_PGOFF_TX_RING ); +if (tx_map ==MAP_FAILED ) { err = -errno; goto out_mmap_rx; } tx->mask = xsk->config.tx_size -1 ; tx->size = xsk->config.tx_size; - tx->producer = map + off.tx.producer; - tx->consumer = map + off.tx.consumer; - tx->ring = map + off.tx.desc; + tx->producer = tx_map + off.tx.producer; + tx->consumer = tx_map + off.tx.consumer; + tx->ring = tx_map + off.tx.desc; tx->cached_cons = xsk->config.tx_size; } xsk->tx = tx;
@@ -653,13 +652,11 @@ int xsk_socket__create(struct xsk_socket **xsk_ptr, const char *ifname, out_mmap_tx:if (tx) -munmap (xsk->tx, - off.tx.desc + +munmap (tx_map, off.tx.desc + xsk->config.tx_size *sizeof (struct xdp_desc )); out_mmap_rx:if (rx) -munmap (xsk->rx, - off.rx.desc + +munmap (rx_map, off.rx.desc + xsk->config.rx_size *sizeof (struct xdp_desc )); out_socket:if (--umem->refcount)
@@ -684,10 +681,12 @@ int xsk_umem__delete(struct xsk_umem *umem) optlen =sizeof (off); err =getsockopt (umem->fd,SOL_XDP ,XDP_MMAP_OFFSETS , &off, &optlen);if (!err) { -munmap (umem->fill->ring, - off.fr.desc + umem->config.fill_size *sizeof (__u64)); -munmap (umem->comp->ring, - off.cr.desc + umem->config.comp_size *sizeof (__u64)); + (void )munmap (umem->fill->ring - off.fr.desc, + off.fr.desc + + umem->config.fill_size *sizeof (__u64)); + (void )munmap (umem->comp->ring - off.cr.desc, + off.cr.desc + + umem->config.comp_size *sizeof (__u64)); }close (umem->fd);
@@ -698,6 +697,7 @@ int xsk_umem__delete(struct xsk_umem *umem)void xsk_socket__delete (struct xsk_socket *xsk) { +size_t desc_sz =sizeof (struct xdp_desc );struct xdp_mmap_offsets off;socklen_t optlen;int err;
@@ -710,14 +710,17 @@ void xsk_socket__delete(struct xsk_socket *xsk) optlen =sizeof (off); err =getsockopt (xsk->fd,SOL_XDP ,XDP_MMAP_OFFSETS , &off, &optlen);if (!err) { -if (xsk->rx) -munmap (xsk->rx->ring, - off.rx.desc + - xsk->config.rx_size *sizeof (struct xdp_desc )); -if (xsk->tx) -munmap (xsk->tx->ring, - off.tx.desc + - xsk->config.tx_size *sizeof (struct xdp_desc )); +if (xsk->rx) { + (void )munmap (xsk->rx->ring - off.rx.desc, + off.rx.desc + + xsk->config.rx_size * desc_sz); + } +if (xsk->tx) { + (void )munmap (xsk->tx->ring - off.tx.desc, + off.tx.desc + + xsk->config.tx_size * desc_sz); + } + } xsk->umem->refcount--;
Signature · 1 line
2.20.1
Re: [PATCH bpf 1/2] libbpf: fix invalid munmap call
Quoted 8 lines · Björn Töpel
From: Björn Töpel <[email protected]>
When unmapping the
AF_XDPmemory regions used for the rings, an invalid address was passed to themunmap()calls. Instead of passing the beginning of the memory region, the descriptor region was passed to munmap.
Quoted 8 lines
When the userspace application tried to tear down an
AF_XDPsocket, the operation failed and the application would still have a reference to socket it wished to get rid of.Reported-by William Tu <[email protected]>Fixes 1cad07884239 ("libbpf: add support for using AF_XDP sockets")Signed-off-by Björn Töpel <[email protected]>---
Thank you.
Daniel Borkmann
maintainerRe: [PATCH bpf 1/2] libbpf: fix invalid munmap call
In reply to Björn TöpelQuoted 14 lines · Björn Töpel
From: Björn Töpel <[email protected]>
When unmapping the
AF_XDPmemory regions used for the rings, an invalid address was passed to themunmap()calls. Instead of passing the beginning of the memory region, the descriptor region was passed to munmap.When the userspace application tried to tear down an
AF_XDPsocket, the operation failed and the application would still have a reference to socket it wished to get rid of.Reported-by William Tu <[email protected]>Fixes 1cad07884239 ("libbpf: add support for using AF_XDP sockets")Signed-off-by Björn Töpel <[email protected]>
[...]
Quoted 18 lines
out_mmap_tx:tools/lib/bpf/xsk.c +2
-4if (tx) -munmap (xsk->tx, - off.tx.desc + +munmap (tx_map, off.tx.desc + xsk->config.tx_size *sizeof (struct xdp_desc )); out_mmap_rx:if (rx) -munmap (xsk->rx, - off.rx.desc + +munmap (rx_map, off.rx.desc + xsk->config.rx_size *sizeof (struct xdp_desc )); out_socket:if (--umem->refcount)@@ -684,10 +681,12 @@ int xsk_umem__delete(struct xsk_umem *umem) optlen =sizeof (off); err =getsockopt (umem->fd,SOL_XDP ,XDP_MMAP_OFFSETS , &off, &optlen);if (!err) {
tools/lib/bpf/xsk.c int xsk_umem__delete(struct xsk_umem *umem) · line 689
Quoted 10 lines
tools/lib/bpf/xsk.c +6
-4-munmap (umem->fill->ring, - off.fr.desc + umem->config.fill_size *sizeof (__u64)); -munmap (umem->comp->ring, - off.cr.desc + umem->config.comp_size *sizeof (__u64)); + (void )munmap (umem->fill->ring - off.fr.desc, + off.fr.desc + + umem->config.fill_size *sizeof (__u64)); + (void )munmap (umem->comp->ring - off.cr.desc, + off.cr.desc + + umem->config.comp_size *sizeof (__u64));
What's the rationale to cast to void here and other places (e.g. below)? If there's no proper reason, then lets remove it. Given the patch has already been applied, please send a follow up. Thanks.
Quoted 38 lines
tools/lib/bpf/xsk.c +12
-8}close (umem->fd);@@ -698,6 +697,7 @@ int xsk_umem__delete(struct xsk_umem *umem)void xsk_socket__delete (struct xsk_socket *xsk) { +size_t desc_sz =sizeof (struct xdp_desc );struct xdp_mmap_offsets off;socklen_t optlen;int err;@@ -710,14 +710,17 @@ void xsk_socket__delete(struct xsk_socket *xsk) optlen =sizeof (off); err =getsockopt (xsk->fd,SOL_XDP ,XDP_MMAP_OFFSETS , &off, &optlen);if (!err) { -if (xsk->rx) -munmap (xsk->rx->ring, - off.rx.desc + - xsk->config.rx_size *sizeof (struct xdp_desc )); -if (xsk->tx) -munmap (xsk->tx->ring, - off.tx.desc + - xsk->config.tx_size *sizeof (struct xdp_desc )); +if (xsk->rx) { + (void )munmap (xsk->rx->ring - off.rx.desc, + off.rx.desc + + xsk->config.rx_size * desc_sz); + } +if (xsk->tx) { + (void )munmap (xsk->tx->ring - off.tx.desc, + off.tx.desc + + xsk->config.tx_size * desc_sz); + } + } xsk->umem->refcount--;
Re: [PATCH bpf 1/2] libbpf: fix invalid munmap call
Quoted 41 lines · Daniel Borkmann
Quoted 14 lines · Björn Töpel
From: Björn Töpel <[email protected]>
When unmapping the
AF_XDPmemory regions used for the rings, an invalid address was passed to themunmap()calls. Instead of passing the beginning of the memory region, the descriptor region was passed to munmap.When the userspace application tried to tear down an
AF_XDPsocket, the operation failed and the application would still have a reference to socket it wished to get rid of.Reported-by William Tu <[email protected]>Fixes 1cad07884239 ("libbpf: add support for using AF_XDP sockets")Signed-off-by Björn Töpel <[email protected]>[...]
Quoted 24 lines
out_mmap_tx:if (tx)tools/lib/bpf/xsk.c +4
-8-munmap (xsk->tx, - off.tx.desc + +munmap (tx_map, off.tx.desc + xsk->config.tx_size *sizeof (struct xdp_desc )); out_mmap_rx:if (rx) -munmap (xsk->rx, - off.rx.desc + +munmap (rx_map, off.rx.desc + xsk->config.rx_size *sizeof (struct xdp_desc )); out_socket:if (--umem->refcount)@@ -684,10 +681,12 @@ int xsk_umem__delete(struct xsk_umem *umem) optlen =sizeof (off); err =getsockopt (umem->fd,SOL_XDP ,XDP_MMAP_OFFSETS , &off, &optlen);if (!err) { -munmap (umem->fill->ring, - off.fr.desc + umem->config.fill_size *sizeof (__u64)); -munmap (umem->comp->ring, - off.cr.desc + umem->config.comp_size *sizeof (__u64)); + (void)munmap(umem->fill->ring - off.fr.desc, + off.fr.desc +
Quoted 9 lines
tools/lib/bpf/xsk.c int xsk_umem__delete(struct xsk_umem *umem) · line 689
Quoted 4 lines
tools/lib/bpf/xsk.c +4
-0+ umem->config.fill_size * sizeof(__u64)); + (void )munmap (umem->comp->ring - off.cr.desc, + off.cr.desc + + umem->config.comp_size *sizeof (__u64));What's the rationale to cast to void here and other places (e.g. below)? If there's no proper reason, then lets remove it. Given the patch has already been applied, please send a follow up. Thanks.
The rationale was to make it explicit that the return value is *not* cared about. If this is not common practice, I'll remove it in a follow up!
Thank,
Björn
Quoted 39 lines
Quoted 38 lines
tools/lib/bpf/xsk.c +12
-8}close (umem->fd);@@ -698,6 +697,7 @@ int xsk_umem__delete(struct xsk_umem *umem)void xsk_socket__delete (struct xsk_socket *xsk) { +size_t desc_sz =sizeof (struct xdp_desc );struct xdp_mmap_offsets off;socklen_t optlen;int err;@@ -710,14 +710,17 @@ void xsk_socket__delete(struct xsk_socket *xsk) optlen =sizeof (off); err =getsockopt (xsk->fd,SOL_XDP ,XDP_MMAP_OFFSETS , &off, &optlen);if (!err) { -if (xsk->rx) -munmap (xsk->rx->ring, - off.rx.desc + - xsk->config.rx_size *sizeof (struct xdp_desc )); -if (xsk->tx) -munmap (xsk->tx->ring, - off.tx.desc + - xsk->config.tx_size *sizeof (struct xdp_desc )); +if (xsk->rx) { + (void )munmap (xsk->rx->ring - off.rx.desc, + off.rx.desc + + xsk->config.rx_size * desc_sz); + } +if (xsk->tx) { + (void )munmap (xsk->tx->ring - off.tx.desc, + off.tx.desc + + xsk->config.tx_size * desc_sz); + } + } xsk->umem->refcount--;
Daniel Borkmann
maintainerRe: [PATCH bpf 1/2] libbpf: fix invalid munmap call
Quoted 50 lines · Björn Töpel
Quoted 40 lines · Daniel Borkmann
Quoted 14 lines · Björn Töpel
From: Björn Töpel <[email protected]>
When unmapping the
AF_XDPmemory regions used for the rings, an invalid address was passed to themunmap()calls. Instead of passing the beginning of the memory region, the descriptor region was passed to munmap.When the userspace application tried to tear down an
AF_XDPsocket, the operation failed and the application would still have a reference to socket it wished to get rid of.Reported-by William Tu <[email protected]>Fixes 1cad07884239 ("libbpf: add support for using AF_XDP sockets")Signed-off-by Björn Töpel <[email protected]>[...]
Quoted 24 lines
out_mmap_tx:if (tx)tools/lib/bpf/xsk.c +4
-8-munmap (xsk->tx, - off.tx.desc + +munmap (tx_map, off.tx.desc + xsk->config.tx_size *sizeof (struct xdp_desc )); out_mmap_rx:if (rx) -munmap (xsk->rx, - off.rx.desc + +munmap (rx_map, off.rx.desc + xsk->config.rx_size *sizeof (struct xdp_desc )); out_socket:if (--umem->refcount)@@ -684,10 +681,12 @@ int xsk_umem__delete(struct xsk_umem *umem) optlen =sizeof (off); err =getsockopt (umem->fd,SOL_XDP ,XDP_MMAP_OFFSETS , &off, &optlen);if (!err) { -munmap (umem->fill->ring, - off.fr.desc + umem->config.fill_size *sizeof (__u64)); -munmap (umem->comp->ring, - off.cr.desc + umem->config.comp_size *sizeof (__u64)); + (void)munmap(umem->fill->ring - off.fr.desc, + off.fr.desc +Quoted 8 lines
tools/lib/bpf/xsk.c int xsk_umem__delete(struct xsk_umem *umem) · line 689
Quoted 4 lines
tools/lib/bpf/xsk.c +4
-0+ umem->config.fill_size * sizeof(__u64)); + (void )munmap (umem->comp->ring - off.cr.desc, + off.cr.desc + + umem->config.comp_size *sizeof (__u64));What's the rationale to cast to void here and other places (e.g. below)? If there's no proper reason, then lets remove it. Given the patch has already been applied, please send a follow up. Thanks.
Quoted 3 lines
The rationale was to make it explicit that the return value is *not* cared about. If this is not common practice, I'll remove it in a follow up!
Yeah, it's not common practice, so lets rather remove it.
Thanks,
Daniel
PATCH 2/2 libbpf: proper XSKMAP cleanup
In reply to Björn TöpelFrom: Björn Töpel <[email protected]>
The bpf_map_update_elem() function, when used on an XSKMAP, will fail if not a valid AF_XDP socket is passed as value. Therefore, this is function cannot be used to clear the XSKMAP. Instead, the bpf_map_delete_elem() function should be used for that.
This patch also simplifies the code by breaking up xsk_update_bpf_maps() into three smaller functions.
--- tools/lib/bpf/xsk.c | 115 +++++++++++++++++++++++--------------------- 1 file changed, 60 insertions(+), 55 deletions(-)
tools/lib/bpf/xsk.c +60 -55
@@ -386,21 +386,17 @@ static void xsk_delete_bpf_maps(struct xsk_socket *xsk) {close (xsk->qidconf_map_fd);close (xsk->xsks_map_fd); + xsk->qidconf_map_fd =-1 ; + xsk->xsks_map_fd =-1 ; } -static int xsk_update_bpf_maps (struct xsk_socket *xsk,int qidconf_value, -int xsks_value) +static int xsk_lookup_bpf_maps (struct xsk_socket *xsk) { -bool qidconf_map_updated = false, xsks_map_updated = false; +__u32 i, *map_ids, num_maps, prog_len =sizeof (struct bpf_prog_info ); +__u32 map_len =sizeof (struct bpf_map_info );struct bpf_prog_info prog_info = {}; -__u32 prog_len =sizeof (prog_info);struct bpf_map_info map_info; -__u32 map_len =sizeof (map_info); -__u32 *map_ids; -int reset_value =0 ; -__u32 num_maps; -unsigned int i; -int err; +int fd, err; err =bpf_obj_get_info_by_fd (xsk->prog_fd, &prog_info, &prog_len);if (err)
@@ -421,66 +417,71 @@ static int xsk_update_bpf_maps(struct xsk_socket *xsk, int qidconf_value, goto out_map_ids;for (i =0 ; i < prog_info.nr_map_ids; i++) { -int fd; +if (xsk->qidconf_map_fd !=-1 && xsk->xsks_map_fd !=-1 ) +break ; fd =bpf_map_get_fd_by_id (map_ids[i]); -if (fd <0 ) { - err = -errno; - goto out_maps; - } +if (fd <0 ) +continue ; err =bpf_obj_get_info_by_fd (fd, &map_info, &map_len); -if (err) - goto out_maps; +if (err) { +close (fd); +continue ; + }if (!strcmp (map_info.name,"qidconf_map" )) { - err =bpf_map_update_elem (fd, &xsk->queue_id, - &qidconf_value,0 ); -if (err) - goto out_maps; - qidconf_map_updated = true; xsk->qidconf_map_fd = fd; - }else if (!strcmp (map_info.name,"xsks_map" )) { - err =bpf_map_update_elem (fd, &xsk->queue_id, - &xsks_value,0 ); -if (err) - goto out_maps; - xsks_map_updated = true; +continue ; + } + +if (!strcmp (map_info.name,"xsks_map" )) { xsk->xsks_map_fd = fd; +continue ; } -if (qidconf_map_updated && xsks_map_updated) -break ; +close (fd); } -if (!(qidconf_map_updated && xsks_map_updated)) { + err =0 ; +if (xsk->qidconf_map_fd <0 || xsk->xsks_map_fd <0 ) { err = -ENOENT ; - goto out_maps; +xsk_delete_bpf_maps (xsk); } - err =0 ; - goto out_success; - -out_maps: -if (qidconf_map_updated) - (void )bpf_map_update_elem (xsk->qidconf_map_fd, &xsk->queue_id, - &reset_value,0 ); -if (xsks_map_updated) - (void )bpf_map_update_elem (xsk->xsks_map_fd, &xsk->queue_id, - &reset_value,0 ); -out_success: -if (qidconf_map_updated) -close (xsk->qidconf_map_fd); -if (xsks_map_updated) -close (xsk->xsks_map_fd); out_map_ids:free (map_ids);return err; } +static void xsk_clear_bpf_maps (struct xsk_socket *xsk) +{ +int qid = false; + + (void )bpf_map_update_elem (xsk->qidconf_map_fd, &xsk->queue_id, &qid,0 ); + (void )bpf_map_delete_elem (xsk->xsks_map_fd, &xsk->queue_id); +} + +static int xsk_set_bpf_maps (struct xsk_socket *xsk) +{ +int qid = true, fd = xsk->fd, err; + + err =bpf_map_update_elem (xsk->qidconf_map_fd, &xsk->queue_id, &qid,0 ); +if (err) + goto out; + + err =bpf_map_update_elem (xsk->xsks_map_fd, &xsk->queue_id, &fd,0 ); +if (err) + goto out; + +return 0 ; +out: +xsk_clear_bpf_maps (xsk); +return err; +} +static int xsk_setup_xdp_prog (struct xsk_socket *xsk) { -bool prog_attached = false;__u32 prog_id =0 ;int err;
@@ -490,7 +491,6 @@ static int xsk_setup_xdp_prog(struct xsk_socket *xsk)return err;if (!prog_id) { - prog_attached = true; err =xsk_create_bpf_maps (xsk);if (err)return err;
@@ -500,20 +500,21 @@ static int xsk_setup_xdp_prog(struct xsk_socket *xsk) goto out_maps; }else { xsk->prog_fd =bpf_prog_get_fd_by_id (prog_id); + err =xsk_lookup_bpf_maps (xsk); +if (err) + goto out_load; } - err =xsk_update_bpf_maps (xsk, true, xsk->fd); + err =xsk_set_bpf_maps (xsk);if (err) goto out_load;return 0 ; out_load: -if (prog_attached) -close (xsk->prog_fd); +close (xsk->prog_fd); out_maps: -if (prog_attached) -xsk_delete_bpf_maps (xsk); +xsk_delete_bpf_maps (xsk);return err; }
@@ -641,6 +642,9 @@ int xsk_socket__create(struct xsk_socket **xsk_ptr, const char *ifname, goto out_mmap_tx; } + xsk->qidconf_map_fd =-1 ; + xsk->xsks_map_fd =-1 ; +if (!(xsk->config.libbpf_flags &XSK_LIBBPF_FLAGS__INHIBIT_PROG_LOAD )) { err =xsk_setup_xdp_prog (xsk);if (err)
@@ -705,7 +709,8 @@ void xsk_socket__delete(struct xsk_socket *xsk)if (!xsk)return ; - (void )xsk_update_bpf_maps (xsk,0 ,0 ); +xsk_clear_bpf_maps (xsk); +xsk_delete_bpf_maps (xsk); optlen =sizeof (off); err =getsockopt (xsk->fd,SOL_XDP ,XDP_MMAP_OFFSETS , &off, &optlen);
Signature · 1 line
2.20.1
Re: [PATCH bpf 2/2] libbpf: proper XSKMAP cleanup
Quoted 8 lines · Björn Töpel
From: Björn Töpel <[email protected]>
The
bpf_map_update_elem()function, when used on an XSKMAP, will fail if not a validAF_XDPsocket is passed as value. Therefore, this is function cannot be used to clear the XSKMAP. Instead, thebpf_map_delete_elem()function should be used for that.
Quoted 7 lines
This patch also simplifies the code by breaking up
xsk_update_bpf_maps()into three smaller functions.Reported-by William Tu <[email protected]>Fixes 1cad07884239 ("libbpf: add support for using AF_XDP sockets")Signed-off-by Björn Töpel <[email protected]>---
Thank you.
Quoted 203 lines
tools/lib/bpf/xsk.c | 115 +++++++++++++++++++++++--------------------- 1 file changed, 60 insertions(+), 55 deletions(-)tools/lib/bpf/xsk.c +60
-56@@ -386,21 +386,17 @@ static void xsk_delete_bpf_maps(struct xsk_socket *xsk) {close (xsk->qidconf_map_fd);close (xsk->xsks_map_fd); + xsk->qidconf_map_fd =-1 ; + xsk->xsks_map_fd =-1 ; } -static int xsk_update_bpf_maps (struct xsk_socket *xsk,int qidconf_value, -int xsks_value) +static int xsk_lookup_bpf_maps (struct xsk_socket *xsk) { -bool qidconf_map_updated = false, xsks_map_updated = false; +__u32 i, *map_ids, num_maps, prog_len =sizeof (struct bpf_prog_info ); +__u32 map_len =sizeof (struct bpf_map_info );struct bpf_prog_info prog_info = {}; -__u32 prog_len =sizeof (prog_info);struct bpf_map_info map_info; -__u32 map_len =sizeof (map_info); -__u32 *map_ids; -int reset_value =0 ; -__u32 num_maps; -unsigned int i; -int err; +int fd, err; err =bpf_obj_get_info_by_fd (xsk->prog_fd, &prog_info, &prog_len);if (err)@@ -421,66 +417,71 @@ static int xsk_update_bpf_maps(struct xsk_socket *xsk, int qidconf_value, goto out_map_ids;for (i =0 ; i < prog_info.nr_map_ids; i++) { -int fd; +if (xsk->qidconf_map_fd !=-1 && xsk->xsks_map_fd !=-1 ) +break ; fd =bpf_map_get_fd_by_id (map_ids[i]); -if (fd <0 ) { - err = -errno; - goto out_maps; - } +if (fd <0 ) +continue ; err =bpf_obj_get_info_by_fd (fd, &map_info, &map_len); -if (err) - goto out_maps; +if (err) { +close (fd); +continue ; + }if (!strcmp (map_info.name,"qidconf_map" )) { - err =bpf_map_update_elem (fd, &xsk->queue_id, - &qidconf_value,0 ); -if (err) - goto out_maps; - qidconf_map_updated = true; xsk->qidconf_map_fd = fd; - }else if (!strcmp (map_info.name,"xsks_map" )) { - err =bpf_map_update_elem (fd, &xsk->queue_id, - &xsks_value,0 ); -if (err) - goto out_maps; - xsks_map_updated = true; +continue ; + } + +if (!strcmp (map_info.name,"xsks_map" )) { xsk->xsks_map_fd = fd; +continue ; } -if (qidconf_map_updated && xsks_map_updated) -break ; +close (fd); } -if (!(qidconf_map_updated && xsks_map_updated)) { + err =0 ; +if (xsk->qidconf_map_fd <0 || xsk->xsks_map_fd <0 ) { err = -ENOENT ; - goto out_maps; +xsk_delete_bpf_maps (xsk); } - err =0 ; - goto out_success; - -out_maps: -if (qidconf_map_updated) - (void )bpf_map_update_elem (xsk->qidconf_map_fd, &xsk->queue_id, - &reset_value,0 ); -if (xsks_map_updated) - (void )bpf_map_update_elem (xsk->xsks_map_fd, &xsk->queue_id, - &reset_value,0 ); -out_success: -if (qidconf_map_updated) -close (xsk->qidconf_map_fd); -if (xsks_map_updated) -close (xsk->xsks_map_fd); out_map_ids:free (map_ids);return err; } +static void xsk_clear_bpf_maps (struct xsk_socket *xsk) +{ +int qid = false; + + (void )bpf_map_update_elem (xsk->qidconf_map_fd, &xsk->queue_id, &qid,0 ); + (void )bpf_map_delete_elem (xsk->xsks_map_fd, &xsk->queue_id); +} + +static int xsk_set_bpf_maps (struct xsk_socket *xsk) +{ +int qid = true, fd = xsk->fd, err; + + err =bpf_map_update_elem (xsk->qidconf_map_fd, &xsk->queue_id, &qid,0 ); +if (err) + goto out; + + err =bpf_map_update_elem (xsk->xsks_map_fd, &xsk->queue_id, &fd,0 ); +if (err) + goto out; + +return 0 ; +out: +xsk_clear_bpf_maps (xsk); +return err; +} +static int xsk_setup_xdp_prog (struct xsk_socket *xsk) { -bool prog_attached = false;__u32 prog_id =0 ;int err;@@ -490,7 +491,6 @@ static int xsk_setup_xdp_prog(struct xsk_socket *xsk)return err;if (!prog_id) { - prog_attached = true; err =xsk_create_bpf_maps (xsk);if (err)return err;@@ -500,20 +500,21 @@ static int xsk_setup_xdp_prog(struct xsk_socket *xsk) goto out_maps; }else { xsk->prog_fd =bpf_prog_get_fd_by_id (prog_id); + err =xsk_lookup_bpf_maps (xsk); +if (err) + goto out_load; } - err =xsk_update_bpf_maps (xsk, true, xsk->fd); + err =xsk_set_bpf_maps (xsk);if (err) goto out_load;return 0 ; out_load: -if (prog_attached) -close (xsk->prog_fd); +close (xsk->prog_fd); out_maps: -if (prog_attached) -xsk_delete_bpf_maps (xsk); +xsk_delete_bpf_maps (xsk);return err; }@@ -641,6 +642,9 @@ int xsk_socket__create(struct xsk_socket **xsk_ptr, const char *ifname, goto out_mmap_tx; } + xsk->qidconf_map_fd =-1 ; + xsk->xsks_map_fd =-1 ; +if (!(xsk->config.libbpf_flags &XSK_LIBBPF_FLAGS__INHIBIT_PROG_LOAD )) { err =xsk_setup_xdp_prog (xsk);if (err)@@ -705,7 +709,8 @@ void xsk_socket__delete(struct xsk_socket *xsk)if (!xsk)return ; - (void )xsk_update_bpf_maps (xsk,0 ,0 ); +xsk_clear_bpf_maps (xsk); +xsk_delete_bpf_maps (xsk); optlen =sizeof (off); err =getsockopt (xsk->fd,SOL_XDP ,XDP_MMAP_OFFSETS , &off, &optlen); --2.20.1
Quoted 11 lines · Björn Töpel
William found two bugs, when doing socket teardown within the same process.
The first issue was an invalid munmap call, and the second one was an invalid XSKMAP cleanup. Both resulted in that the process kept references to the socket, which was not correctly cleaned up. When a new socket was created, the
bind()call would fail, since the old socket was still lingering, refusing to give up the queue on the netdev.More details can be found in the individual commits.
Alexei Starovoitov
AppliedQuoted 15 lines · Jonathan Lemon
Quoted 10 lines · Björn Töpel
William found two bugs, when doing socket teardown within the same process.
The first issue was an invalid munmap call, and the second one was an invalid XSKMAP cleanup. Both resulted in that the process kept references to the socket, which was not correctly cleaned up. When a new socket was created, the
bind()call would fail, since the old socket was still lingering, refusing to give up the queue on the netdev.
Quoted 3 lines
Quoted 1 line
More details can be found in the individual commits.
Reviewed-by Jonathan Lemon <[email protected]>
Applied. Thanks!