- 22 May, 2022 25 commits
-
-
David Howells authored
If a client's address changes, say if it is NAT'd, this can disrupt an in progress operation. For most operations, this is not much of a problem, but StoreData can be different as some servers modify the target file as the data comes in, so if a store request is disrupted, the file can get corrupted on the server. The problem is that the server doesn't recognise packets that come after the change of address as belonging to the original client and will bounce them, either by sending an OUT_OF_SEQUENCE ACK to the apparent new call if the packet number falls within the initial sequence number window of a call or by sending an EXCEEDS_WINDOW ACK if it falls outside and then aborting it. In both cases, firstPacket will be 1 and previousPacket will be 0 in the ACK information. Fix this by the following means: (1) If a client call receives an EXCEEDS_WINDOW ACK with firstPacket as 1 and previousPacket as 0, assume this indicates that the server saw the incoming packets from a different peer and thus as a different call. Fail the call with error -ENETRESET. (2) Also fail the call if a similar OUT_OF_SEQUENCE ACK occurs if the first packet has been hard-ACK'd. If it hasn't been hard-ACK'd, the ACK packet will cause it to get retransmitted, so the call will just be repeated. (3) Make afs_select_fileserver() treat -ENETRESET as a straight fail of the operation. (4) Prioritise the error code over things like -ECONNRESET as the server did actually respond. (5) Make writeback treat -ENETRESET as a retryable error and make it redirty all the pages involved in a write so that the VM will retry. Note that there is still a circumstance that I can't easily deal with: if the operation is fully received and processed by the server, but the reply is lost due to address change. There's no way to know if the op happened. We can examine the server, but a conflicting change could have been made by a third party - and we can't tell the difference. In such a case, a message like: kAFS: vnode modified {100058:146266} b7->b8 YFS.StoreData64 (op=2646a) will be logged to dmesg on the next op to touch the file and the client will reset the inode state, including invalidating clean parts of the pagecache. Reported-by:Marc Dionne <marc.dionne@auristor.com> Signed-off-by:
David Howells <dhowells@redhat.com> cc: linux-afs@lists.infradead.org Link: http://lists.infradead.org/pipermail/linux-afs/2021-December/004811.html # v1 Signed-off-by:
David S. Miller <davem@davemloft.net>
-
David Howells authored
The RX_USER_ABORT code should really only be used to indicate that the user of the rxrpc service (ie. userspace) implicitly caused a call to be aborted - for instance if the AF_RXRPC socket is closed whilst the call was in progress. (The user may also explicitly abort a call and specify the abort code to use). Change some of the points of generation to use other abort codes instead: (1) Abort the call with RXGEN_SS_UNMARSHAL or RXGEN_CC_UNMARSHAL if we see ENOMEM and EFAULT during received data delivery and abort with RX_CALL_DEAD in the default case. (2) Abort with RXGEN_SS_MARSHAL if we get ENOMEM whilst trying to send a reply. (3) Abort with RX_CALL_DEAD if we stop hearing from the peer if we had heard from the peer and abort with RX_CALL_TIMEOUT if we hadn't. (4) Abort with RX_CALL_DEAD if we try to disconnect a call that's not completed successfully or been aborted. Reported-by:Jeffrey Altman <jaltman@auristor.com> Signed-off-by:
-
David Howells authored
If at the end of rxrpc sendmsg() or rxrpc_kernel_send_data() the call that was being given data was aborted remotely or otherwise failed, return an error rather than returning the amount of data buffered for transmission. The call (presumably) did not complete, so there's not much point continuing with it. AF_RXRPC considers it "complete" and so will be unwilling to do anything else with it - and won't send a notification for it, deeming the return from sendmsg sufficient. Not returning an error causes afs to incorrectly handle a StoreData operation that gets interrupted by a change of address due to NAT reconfiguration. This doesn't normally affect most operations since their request parameters tend to fit into a single UDP packet and afs_make_call() returns before the server responds; StoreData is different as it involves transmission of a lot of data. This can be triggered on a client by doing something like: dd if=/dev/zero of=/afs/example.com/foo bs=1M count=512 at one prompt, and then changing the network address at another prompt, e.g.: ifconfig enp6s0 inet 192.168.6.2 && route add 192.168.6.1 dev enp6s0 Tracing packets on an Auristor fileserver looks something like: 192.168.6.1 -> 192.168.6.3 RX 107 ACK Idle Seq: 0 Call: 4 Source Port: 7000 Destination Port: 7001 192.168.6.3 -> 192.168.6.1 AFS (RX) 1482 FS Request: Unknown(64538) (64538) 192.168.6.3 -> 192.168.6.1 AFS (RX) 1482 FS Request: Unknown(64538) (64538) 192.168.6.1 -> 192.168.6.3 RX 107 ACK Idle Seq: 0 Call: 4 Source Port: 7000 Destination Port: 7001 <ARP exchange for 192.168.6.2> 192.168.6.2 -> 192.168.6.1 AFS (RX) 1482 FS Request: Unknown(0) (0) 192.168.6.2 -> 192.168.6.1 AFS (RX) 1482 FS Request: Unknown(0) (0) 192.168.6.1 -> 192.168.6.2 RX 107 ACK Exceeds Window Seq: 0 Call: 4 Source Port: 7000 Destination Port: 7001 192.168.6.1 -> 192.168.6.2 RX 74 ABORT Seq: 0 Call: 4 Source Port: 7000 Destination Port: 7001 192.168.6.1 -> 192.168.6.2 RX 74 ABORT Seq: 29321 Call: 4 Source Port: 7000 Destination Port: 7001 The Auristor fileserver logs code -453 (RXGEN_SS_UNMARSHAL), but the abort code received by kafs is -5 (RX_PROTOCOL_ERROR) as the rx layer sees the condition and generates an abort first and the unmarshal error is a consequence of that at the application layer. Reported-by:
-
David Howells authored
Automatically generate trace tag enums from the symbol -> string mapping tables rather than having the enums as well, thereby reducing duplicated data. Signed-off-by:
-
David Howells authored
There's a locking issue with the per-netns list of calls in rxrpc. The pieces of code that add and remove a call from the list use write_lock() and the calls procfile uses read_lock() to access it. However, the timer callback function may trigger a removal by trying to queue a call for processing and finding that it's already queued - at which point it has a spare refcount that it has to do something with. Unfortunately, if it puts the call and this reduces the refcount to 0, the call will be removed from the list. Unfortunately, since the _bh variants of the locking functions aren't used, this can deadlock. ================================ WARNING: inconsistent lock state 5.18.0-rc3-build4+ #10 Not tainted -------------------------------- inconsistent {SOFTIRQ-ON-W} -> {IN-SOFTIRQ-W} usage. ksoftirqd/2/25 [HC0[0]:SC1[1]:HE1:SE0] takes: ffff888107ac4038 (&rxnet->call_lock){+.?.}-{2:2}, at: rxrpc_put_call+0x103/0x14b {SOFTIRQ-ON-W} state was registered at: ... Possible unsafe locking scenario: CPU0 ---- lock(&rxnet->call_lock); <Interrupt> lock(&rxnet->call_lock); *** DEADLOCK *** 1 lock held by ksoftirqd/2/25: #0: ffff8881008ffdb0 ((&call->timer)){+.-.}-{0:0}, at: call_timer_fn+0x5/0x23d Changes ======= ver #2) - Changed to using list_next_rcu() rather than rcu_dereference() directly. Fixes: 17926a79 ("[AF_RXRPC]: Provide secure RxRPC sockets for use by userspace and kernel both") Signed-off-by: -
David Howells authored
Move to using refcount_t rather than atomic_t for refcounts in rxrpc. Signed-off-by:
-
David Howells authored
Allow the list of in-use local UDP endpoints in the current network namespace to be viewed in /proc. To aid with this, the endpoint list is converted to an hlist and RCU-safe manipulation is used so that the list can be read with only the RCU read lock held. Signed-off-by:
-
David S. Miller authored
Alex Elder says: ==================== net: ipa: a few more small items This series consists of three small sets of changes. Version 2 adds a patch that avoids a warning that occurs when handling a modem crash (I unfortunately didn't notice it earlier). All other patches are the same--just rebased. The first three patches allow a few endpoint features to be specified. At this time, currently-defined endpoints retain the same configuration, but when the monitor functionality is added in the next cycle these options will be required. The fourth patch simply removes an unused function, explaining also why it would likely never be used. The fifth patch is new. It counts the number of modem TX endpoints and uses it to determine how many TREs a transaction needs when when handling a modem crash. It is needed to avoid exceeding the limited number of commands imposed by the last four patches. And the last four patches refactor code related to IPA immediate commands, eliminating an unused field and then simplifying and removing some unneeded code. ==================== Signed-off-by: -
Alex Elder authored
The 64-bit data field in a transaction is not used for commands. And the opcode array is *only* used for commands. They're (currently) the same size; save a little space in the transaction structure by enclosing the two fields in a union. Signed-off-by:
Alex Elder <elder@linaro.org> Signed-off-by:
-
Alex Elder authored
The ipa_cmd_info structure now contains only one field, and it's an enumerated type whose values all fit in 8 bits. Currently we'll never use more than 8 TREs in a command transaction, and we can represent that number of command opcodes in the same space as a 64 bit pointer to an ipa_cmd_info structure. Define IPA_COMMAND_TRANS_TRE_MAX as the maximum number of TREs that can be in a command transaction. Replace the info pointer in a transaction with a fixed-size array named cmd_opcode[] of that many bytes. Store the opcode in this array when adding a command TRE to a transaction, as was done previously for the info array. This makes the ipa_cmd_info unused, so get rid of it. When committing an immediate command transaction, use the channel's Boolean command flag to determine whether to fill in the opcode, which will be taken (as before) from the array in the transaction. This makes the command info pool unnecessary, so get rid of it. Signed-off-by:
-
Alex Elder authored
We no longer use the direction argument for gsi_trans_cmd_add(), so get rid of it in its definition, and in its seven callers. Signed-off-by:
-
Alex Elder authored
The direction field of the ipa_cmd_info structure is set, but never used. It seems it might have been used for the DMA_SHARED_MEM immediate command, but the DIRECTION flag is set based on the value of the passed-in direction flag there. Anyway, remove this unused field from the ipa_cmd_info structure. This is done as a separate patch to make it very obvious that it's not required. Signed-off-by:
-
Alex Elder authored
In ipa_endpoint_modem_exception_reset_all(), a high estimate was made of the number of endpoints that need their status register updated. We only used what was needed, so the high estimate didn't matter much. However the next few patches are going to limit the number of commands in a single transaction, and the overestimate would exceed that. So count the number of modem TX endpoints at initialization time, and use it in ipa_endpoint_modem_exception_reset_all(). Signed-off-by:
-
Alex Elder authored
Since the beginning gsi_trans_commit_wait_timeout() has existed to provide a way to allow waiting a limited time for a transaction to complete. But that function has never been used. In fact, there is no use for this function, because a transaction committed to hardware should *always* complete. The only reason it might not complete is if there were a hardware failure, or perhaps a system configuration error. Furthermore, if a timeout ever did occur, the IPA hardware would be in an indeterminate state, from which there is no recovery. It would require some sort of complete IPA reset, and would require the participation of the modem, and at this time there is no such sequence defined. So get rid of the definition of gsi_trans_commit_wait_timeout(), and update a few comments accordingly. Signed-off-by:
-
Alex Elder authored
Don't assume that a 500 microsecond time limit should be used for all receive endpoints that support aggregation. Instead, specify the time limit to use in the configuration data. Set a 500 microsecond limit for all existing RX endpoints, as before. Checking for overflow for the time limit field is a bit complicated. Rather than duplicate a lot of code in ipa_endpoint_data_valid_one(), call WARN() if any value is found to be too large when encoding it. Signed-off-by:
-
Alex Elder authored
Add a new flag for AP receive endpoints that indicates whether a "hard limit" is used as a criterion for closing aggregation. Add comments explaining the difference between "hard" and "soft" aggregation limits. Pass a flag to ipa_aggr_size_kb() so it computes the proper aggregation size value whether using hard or soft limits. Move that function earlier in "ipa_endpoint.c" so it can be used without a forward-reference. Update ipa_endpoint_data_valid_one() so it validates endpoints whose data indicate a hard aggregation limit is used, and so it reports set aggregation flags for endpoints without aggregation enabled. Signed-off-by:
-
Alex Elder authored
Add a new Boolean flag for RX endpoints defining whether HOLB drop is initially enabled or disabled for the endpoint. All existing AP endpoints should have HOLB drop disabled. Signed-off-by:
-
Julia Lawall authored
Spelling mistakes (triple letters) in comments. Detected with the help of Coccinelle. Signed-off-by:
Julia Lawall <Julia.Lawall@inria.fr> Signed-off-by:
-
Julia Lawall authored
Spelling mistake (triple letters) in comment. Detected with the help of Coccinelle. Signed-off-by:
Niklas Söderlund <niklas.soderlund@corigine.com> Signed-off-by:
-
Julia Lawall authored
Spelling mistake (triple letters) in comment. Detected with the help of Coccinelle. Signed-off-by:
-
Julia Lawall authored
Spelling mistake (triple letters) in comment. Detected with the help of Coccinelle. Signed-off-by:
-
Julia Lawall authored
Spelling mistakes (triple letters) in comments. Detected with the help of Coccinelle. Signed-off-by:
-
Julia Lawall authored
Spelling mistake (triple letters) in comment. Detected with the help of Coccinelle. Signed-off-by:
-
Julia Lawall authored
Spelling mistake (triple letters) in comment. Detected with the help of Coccinelle. Signed-off-by:
-
Julia Lawall authored
Spelling mistake (triple letters) in comment. Detected with the help of Coccinelle. Signed-off-by:
-
- 21 May, 2022 12 commits
-
-
Jakub Kicinski authored
Joanne Koong says: ==================== Add a bhash2 table hashed by port + address This patchset proposes adding a bhash2 table that hashes by port and address. The motivation behind bhash2 is to expedite bind requests in situations where the port has many sockets in its bhash table entry, which makes checking bind conflicts costly especially given that we acquire the table entry spinlock while doing so, which can cause softirq cpu lockups and can prevent new tcp connections. We ran into this problem at Meta where the traffic team binds a large number of IPs to port 443 and the bind() call took a significant amount of time which led to cpu softirq lockups, which caused packet drops and other failures on the machine The patches are as follows: 1/2 - Adds a second bhash table (bhash2) hashed by port and address 2/2 - Adds a test for timing how long an additional bind request takes when the bhash entry is populated When experimentally testing this on a local server for ~24k sockets bound to the port, the results seen were: ipv4: before - 0.002317 seconds with bhash2 - 0.000018 seconds ipv6: before - 0.002431 seconds with bhash2 - 0.000021 seconds ==================== Link: https://lore.kernel.org/r/20220520001834.2247810-1-kuba@kernel.orgSigned-off-by:
Jakub Kicinski <kuba@kernel.org>
-
Joanne Koong authored
This test populates the bhash table for a given port with MAX_THREADS * MAX_CONNECTIONS sockets, and then times how long a bind request on the port takes. When populating the bhash table, we create the sockets and then bind the sockets to the same address and port (SO_REUSEADDR and SO_REUSEPORT are set). When timing how long a bind on the port takes, we bind on a different address without SO_REUSEPORT set. We do not set SO_REUSEPORT because we are interested in the case where the bind request does not go through the tb->fastreuseport path, which is fragile (eg tb->fastreuseport path does not work if binding with a different uid). To run the test locally, I did: * ulimit -n 65535000 * ip addr add 2001:0db8:0:f101::1 dev eth0 * ./bind_bhash_test 443 Signed-off-by:
Joanne Koong <joannelkoong@gmail.com> Signed-off-by:
-
Joanne Koong authored
We currently have one tcp bind table (bhash) which hashes by port number only. In the socket bind path, we check for bind conflicts by traversing the specified port's inet_bind2_bucket while holding the bucket's spinlock (see inet_csk_get_port() and inet_csk_bind_conflict()). In instances where there are tons of sockets hashed to the same port at different addresses, checking for a bind conflict is time-intensive and can cause softirq cpu lockups, as well as stops new tcp connections since __inet_inherit_port() also contests for the spinlock. This patch proposes adding a second bind table, bhash2, that hashes by port and ip address. Searching the bhash2 table leads to significantly faster conflict resolution and less time holding the spinlock. Signed-off-by:
Eric Dumazet <edumazet@google.com> Acked-by:
Kuniyuki Iwashima <kuniyu@amazon.co.jp> Signed-off-by:
-
Jakub Kicinski authored
GCC array-bounds warns that ipc_coredump_get_list() under-allocates the size of struct iosm_cd_table *cd_table. This is avoidable - we just need a flexible array. Nothing calls sizeof() on struct iosm_cd_list or anything that contains it. Reviewed-by:
M Chetan Kumar <m.chetan.kumar@intel.com> Link: https://lore.kernel.org/r/20220520060013.2309497-1-kuba@kernel.orgSigned-off-by:
-
Xiu Jianfeng authored
Use memset_after() helper to simplify the code, there is no functional change in this patch. Signed-off-by:
Xiu Jianfeng <xiujianfeng@huawei.com> Link: https://lore.kernel.org/r/20220519062932.249926-1-xiujianfeng@huawei.comSigned-off-by:
-
Alaa Mohamed authored
align the extack argument of the ocelot_port_fdb_del() function. Signed-off-by:
Alaa Mohamed <eng.alaamohamedsoliman.am@gmail.com> Link: https://lore.kernel.org/r/20220520002040.4442-1-eng.alaamohamedsoliman.am@gmail.comSigned-off-by:
-
Alaa Mohamed authored
The continuation line does not align with the opening bracket and this patch fix it. Signed-off-by:
-
Jakub Kicinski authored
GCC array bounds checking complains that ulp_id is validated only against upper bound. Make it unsigned. Reviewed-by:
Michael Chan <michael.chan@broadcom.com> Link: https://lore.kernel.org/r/20220520061955.2312968-1-kuba@kernel.orgSigned-off-by:
-
Amit Cohen authored
Commit 49bb39bd ("selftests: fib_nexthops: Make the test more robust") increased the timeout of ping commands to 5 seconds, to make the test more robust. Make the timeout configurable using '-w' argument to allow user to change it depending on the system that runs the test. Some systems suffer from slow forwarding performance, so they may need to change the timeout. Signed-off-by:
Amit Cohen <amcohen@nvidia.com> Reviewed-by:
Shuah Khan <skhan@linuxfoundation.org> Link: https://lore.kernel.org/r/20220519070921.3559701-1-amcohen@nvidia.comSigned-off-by:
-
Yang Yingliang authored
Sometimes t7xx_cldma_gpd_set_next_ptr() is called under spin lock, so add 'gfp_mask' parameter in t7xx_cldma_gpd_set_next_ptr() to pass the flag. Fixes: 39d43904 ("net: wwan: t7xx: Add control DMA interface") Reported-by:
Hulk Robot <hulkci@huawei.com> Signed-off-by:
Yang Yingliang <yangyingliang@huawei.com> Reviewed-by:
Loic Poulain <loic.poulain@linaro.org> Link: https://lore.kernel.org/r/20220519032108.2996400-1-yangyingliang@huawei.comSigned-off-by:
-
Rolf Eike Beer authored
Fix typo which breaks build for parisc. Fixes: 3daebfbe ("net: tulip: convert to devres") Reported-by:
Linux Kernel Functional Testing <lkft@linaro.org> Link: https://lore.kernel.org/all/CA+G9fYuCzU5VZ_nc+6NEdBXJdVCH=J2SB1Na1G_NS_0BNdGYtg@mail.gmail.com/Reported-by:
kernel test robot <lkp@intel.com> Signed-off-by:
Rolf Eike Beer <eike-kernel@sf-tec.de> Link: https://lore.kernel.org/r/4719560.GXAFRqVoOG@eto.sf-tec.deSigned-off-by:
-
Jakub Kicinski authored
When user sets skb_defer_max to 1 the kick threshold is 0 (half of 1). If we increment queue length before the check the kick will never happen, and the skb may get stranded. This is likely harmless but can be avoided by moving the increment after the check. This way skb_defer_max == 1 will always kick. Still a silly config to have, but somehow that feels more correct. While at it drop a comment which seems to be outdated or confusing, and wrap the defer_count write with a WRITE_ONCE() since it's read on the fast path that avoids taking the lock. Reviewed-by:
-
- 20 May, 2022 3 commits
-
-
Martin Habets authored
Siena only supports software TSO. This means more code can be deleted, as pointed out by the Smatch static checker warning: drivers/net/ethernet/sfc/siena/tx.c:184 __efx_siena_enqueue_skb() warn: duplicate check 'segments' (previous on line 158) Fixes: 956f2d86 ("sfc/siena: Remove build references to missing functionality") Reported-by:
Dan Carpenter <dan.carpenter@oracle.com> Link: https://lore.kernel.org/kernel-janitors/YoH5tJMnwuGTrn1Z@kili/Signed-off-by:
Martin Habets <habetsm.xilinx@gmail.com> Link: https://lore.kernel.org/r/165294463549.23865.4557617334650441347.stgit@palantir17.mph.netSigned-off-by:
-
Jakub Kicinski authored
Looks like the IPv6 version of the patch under Fixes was a copy/paste of the IPv4 but hit the wrong spot. It is tcp_v6_rcv() which uses drop_reason as a boolean, and needs to be protected against reason == 0 before calling free. tcp_v6_do_rcv() has a pretty straightforward flow. The resulting warning looks like this: WARNING: CPU: 1 PID: 0 at net/core/skbuff.c:775 Call Trace: tcp_v6_rcv (net/ipv6/tcp_ipv6.c:1767) ip6_protocol_deliver_rcu (net/ipv6/ip6_input.c:438) ip6_input_finish (include/linux/rcupdate.h:726) ip6_input (include/linux/netfilter.h:307) Fixes: f8319dfd ("net: tcp: reset 'drop_reason' to NOT_SPCIFIED in tcp_v{4,6}_rcv()") Tested-by:Matthieu Baerts <matthieu.baerts@tessares.net> Link: https://lore.kernel.org/r/20220520021347.2270207-1-kuba@kernel.orgSigned-off-by:
-
David S. Miller authored
Alex Elder says: ==================== net: ipa: a mix of patches This series includes a mix of things things that are generally minor. The first four are sort of unrelated fixes, and summarizing them here wouldn't be that helpful. The last three together make it so only the "configuration data" we need after initialization is saved for later use. Most such data is used only during driver initialization. But endpoint configuration is needed later, so the last patch saves a copy of that. Eventually we'll want to support reconfiguring endpoints at runtime as well, and this will facilitate that. ==================== Signed-off-by:
-
