/*
 * netsniff-ng - the packet sniffing beast
 * Copyright 2009, 2010 Daniel Borkmann.
 * Copyright 2009, 2010 Emmanuel Roullit.
 * Subject to the GPL, version 2.
 */

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <sys/mman.h>
#include <sys/types.h>
#include <sys/socket.h>
#include <arpa/inet.h>
#include <linux/if_ether.h>

#include "die.h"
#include "xmalloc.h"
#include "ring_tx.h"
#include "built_in.h"

static void set_packet_loss_discard(int sock)
{
	int ret, discard = 1;
	ret = setsockopt(sock, SOL_PACKET, PACKET_LOSS, (void *) &discard,
			 sizeof(discard));
	if (ret < 0)
		panic("setsockopt: cannot set packet loss");
}

void destroy_tx_ring(int sock, struct ring *ring)
{
	int ret;

	munmap(ring->mm_space, ring->mm_len);
	ring->mm_len = 0;

	fmemset(&ring->layout, 0, sizeof(ring->layout));
	ret = setsockopt(sock, SOL_PACKET, PACKET_TX_RING, &ring->layout,
			 sizeof(ring->layout));
	if (unlikely(ret))
		panic("Cannot destroy the TX_RING: %s!\n", strerror(errno));

	xfree(ring->frames);
}

static void setup_tx_ring_layout(int sock, struct ring *ring, size_t size,
				 bool jumbo_support)
{
	fmemset(&ring->layout, 0, sizeof(ring->layout));

	ring->layout.tp_block_size = (jumbo_support ?
				      RUNTIME_PAGE_SIZE << 4 :
				      RUNTIME_PAGE_SIZE << 2);

	ring->layout.tp_frame_size = (jumbo_support ?
				      TPACKET_ALIGNMENT << 12 :
				      TPACKET_ALIGNMENT << 7);

	ring->layout.tp_block_nr = size / ring->layout.tp_block_size;
	ring->layout.tp_frame_nr = ring->layout.tp_block_size /
				   ring->layout.tp_frame_size *
				   ring->layout.tp_block_nr;

	set_sockopt_tpacket_v2(sock);

	ring_verify_layout(ring);
}

static void create_tx_ring(int sock, struct ring *ring, bool verbose)
{
	int ret;
retry:
	ret = setsockopt(sock, SOL_PACKET, PACKET_TX_RING, &ring->layout,
			 sizeof(ring->layout));

	if (errno == ENOMEM && ring->layout.tp_block_nr > 1) {
		ring->layout.tp_block_nr >>= 1;
		ring->layout.tp_frame_nr = ring->layout.tp_block_size / 
					   ring->layout.tp_frame_size * 
					   ring->layout.tp_block_nr;
		goto retry;
	}

	if (ret < 0)
		panic("Cannot allocate TX_RING!\n");

	ring->mm_len = (size_t) ring->layout.tp_block_size * ring->layout.tp_block_nr;

	if (verbose) {
		printf("TX,V2: %.2Lf MiB, %u Frames, each %u Byte allocated\n",
		       (long double) ring->mm_len / (1 << 20),
		       ring->layout.tp_frame_nr, ring->layout.tp_frame_size);
	}
}

static void alloc_tx_ring_frames(int sock __maybe_unused, struct ring *ring)
{
	alloc_ring_frames_generic(ring, ring->layout.tp_frame_nr,
				  ring->layout.tp_frame_size);
}

void ring_tx_setup(struct ring *ring, int sock, size_t size, int ifindex,
		   bool jumbo_support, bool verbose)
{
	fmemset(ring, 0, sizeof(*ring));
	set_packet_loss_discard(sock);
	setup_tx_ring_layout(sock, ring, size, jumbo_support);
	create_tx_ring(sock, ring, verbose);
	mmap_ring_generic(sock, ring);
	alloc_tx_ring_frames(sock, ring);
	bind_ring_generic(sock, ring, ifindex, true);
}
lass='label'>context:</td><td class='ctrl'><select name='context' onchange='this.form.submit();'><option value='1'>1</option><option value='2'>2</option><option value='3' selected='selected'>3</option><option value='4'>4</option><option value='5'>5</option><option value='6'>6</option><option value='7'>7</option><option value='8'>8</option><option value='9'>9</option><option value='10'>10</option><option value='15'>15</option><option value='20'>20</option><option value='25'>25</option><option value='30'>30</option><option value='35'>35</option><option value='40'>40</option></select></td></tr><tr><td class='label'>space:</td><td class='ctrl'><select name='ignorews' onchange='this.form.submit();'><option value='0' selected='selected'>include</option><option value='1'>ignore</option></select></td></tr><tr><td class='label'>mode:</td><td class='ctrl'><select name='dt' onchange='this.form.submit();'><option value='0' selected='selected'>unified</option><option value='1'>ssdiff</option><option value='2'>stat only</option></select></td></tr><tr><td/><td class='ctrl'><noscript><input type='submit' value='reload'/></noscript></td></tr></table></form></div><table summary='commit info' class='commit-info'>
<tr><th>author</th><td>Ilya Dryomov &lt;idryomov@gmail.com&gt;</td><td class='right'>2016-07-19 03:50:28 +0200</td></tr>
<tr><th>committer</th><td>Ilya Dryomov &lt;idryomov@gmail.com&gt;</td><td class='right'>2016-07-22 15:17:40 +0200</td></tr>
<tr><th>commit</th><td colspan='2' class='oid'><a href='/cgit.cgi/linux/net-next.git/commit/?h=nds-private-remove&amp;id=930c532869774ebf8af9efe9484c597f896a7d46'>930c532869774ebf8af9efe9484c597f896a7d46</a> (<a href='/cgit.cgi/linux/net-next.git/patch/?id=930c532869774ebf8af9efe9484c597f896a7d46'>patch</a>)</td></tr>
<tr><th>tree</th><td colspan='2' class='oid'><a href='/cgit.cgi/linux/net-next.git/tree/?h=nds-private-remove&amp;id=930c532869774ebf8af9efe9484c597f896a7d46'>68072d2a5c42dbf347514b8737f1abd0df044966</a></td></tr>
<tr><th>parent</th><td colspan='2' class='oid'><a href='/cgit.cgi/linux/net-next.git/commit/?h=nds-private-remove&amp;id=92d21ac74a9e3c09b0b01c764e530657e4c85c49'>92d21ac74a9e3c09b0b01c764e530657e4c85c49</a> (<a href='/cgit.cgi/linux/net-next.git/diff/?h=nds-private-remove&amp;id=930c532869774ebf8af9efe9484c597f896a7d46&amp;id2=92d21ac74a9e3c09b0b01c764e530657e4c85c49'>diff</a>)</td></tr></table>
<div class='commit-subject'>libceph: apply new_state before new_up_client on incrementals</div><div class='commit-msg'>Currently, osd_weight and osd_state fields are updated in the encoding
order.  This is wrong, because an incremental map may look like e.g.

    new_up_client: { osd=6, addr=... } # set osd_state and addr
    new_state: { osd=6, xorstate=EXISTS } # clear osd_state

Suppose osd6's current osd_state is EXISTS (i.e. osd6 is down).  After
applying new_up_client, osd_state is changed to EXISTS | UP.  Carrying
on with the new_state update, we flip EXISTS and leave osd6 in a weird
"!EXISTS but UP" state.  A non-existent OSD is considered down by the
mapping code

2087    for (i = 0; i &lt; pg-&gt;pg_temp.len; i++) {
2088            if (ceph_osd_is_down(osdmap, pg-&gt;pg_temp.osds[i])) {
2089                    if (ceph_can_shift_osds(pi))
2090                            continue;
2091
2092                    temp-&gt;osds[temp-&gt;size++] = CRUSH_ITEM_NONE;

and so requests get directed to the second OSD in the set instead of
the first, resulting in OSD-side errors like:

[WRN] : client.4239 192.168.122.21:0/2444980242 misdirected client.4239.1:2827 pg 2.5df899f2 to osd.4 not [1,4,6] in e680/680

and hung rbds on the client:

[  493.566367] rbd: rbd0: write 400000 at 11cc00000 (0)
[  493.566805] rbd: rbd0:   result -6 xferred 400000
[  493.567011] blk_update_request: I/O error, dev rbd0, sector 9330688

The fix is to decouple application from the decoding and:
- apply new_weight first
- apply new_state before new_up_client
- twiddle osd_state flags if marking in
- clear out some of the state if osd is destroyed

Fixes: http://tracker.ceph.com/issues/14901

Cc: stable@vger.kernel.org # 3.15+: 6dd74e44dc1d: libceph: set 'exists' flag for newly up osd
Cc: stable@vger.kernel.org # 3.15+
Signed-off-by: Ilya Dryomov &lt;idryomov@gmail.com&gt;
Reviewed-by: Josh Durgin &lt;jdurgin@redhat.com&gt;
</div><div class='diffstat-header'><a href='/cgit.cgi/linux/net-next.git/diff/?h=nds-private-remove&amp;id=930c532869774ebf8af9efe9484c597f896a7d46'>Diffstat</a></div><table summary='diffstat' class='diffstat'>