linux.git/net/9p, branch v5.15 Linux kernel source tree net/9p: increase default msize to 128k 2021-09-04T23:36:44+00:00 Christian Schoenebeck linux_oss@crudebyte.com 2021-09-04T15:12:51+00:00 9c4d94dc9a64426d2fa0255097a3a84f6ff2eebe Let's raise the default msize value to 128k. The 'msize' option defines the maximum message size allowed for any message being transmitted (in both directions) between 9p server and 9p client during a 9p session. Currently the default 'msize' is just 8k, which is way too conservative. Such a small 'msize' value has quite a negative performance impact, because individual 9p messages have to be split up far too often into numerous smaller messages to fit into this message size limitation. A default value of just 8k also has a much higher probablity of hitting short-read issues like: https://gitlab.com/qemu-project/qemu/-/issues/409 Unfortunately user feedback showed that many 9p users are not aware that this option even exists, nor the negative impact it might have if it is too low. Link: http://lkml.kernel.org/r/61ea0f0faaaaf26dd3c762eabe4420306ced21b9.1630770829.git.linux_oss@crudebyte.com Link: https://lists.gnu.org/archive/html/qemu-devel/2021-03/msg01003.html Signed-off-by: Christian Schoenebeck <linux_oss@crudebyte.com> Signed-off-by: Dominique Martinet <asmadeus@codewreck.org> 
Let's raise the default msize value to 128k.

The 'msize' option defines the maximum message size allowed for any
message being transmitted (in both directions) between 9p server and 9p
client during a 9p session.

Currently the default 'msize' is just 8k, which is way too conservative.
Such a small 'msize' value has quite a negative performance impact,
because individual 9p messages have to be split up far too often into
numerous smaller messages to fit into this message size limitation.

A default value of just 8k also has a much higher probablity of hitting
short-read issues like: https://gitlab.com/qemu-project/qemu/-/issues/409

Unfortunately user feedback showed that many 9p users are not aware that
this option even exists, nor the negative impact it might have if it is
too low.

Link: http://lkml.kernel.org/r/61ea0f0faaaaf26dd3c762eabe4420306ced21b9.1630770829.git.linux_oss@crudebyte.com
Link: https://lists.gnu.org/archive/html/qemu-devel/2021-03/msg01003.html
Signed-off-by: Christian Schoenebeck <linux_oss@crudebyte.com>
Signed-off-by: Dominique Martinet <asmadeus@codewreck.org>
net/9p: use macro to define default msize 2021-09-04T23:36:44+00:00 Christian Schoenebeck linux_oss@crudebyte.com 2021-09-04T15:07:12+00:00 9210fc0a3b61c396ba9b32467b88c89d41bc9fe7 Use a macro to define the default value for the 'msize' option at one place instead of using two separate integer literals. Link: http://lkml.kernel.org/r/28bb651ae0349a7d57e8ddc92c1bd5e62924a912.1630770829.git.linux_oss@crudebyte.com Signed-off-by: Christian Schoenebeck <linux_oss@crudebyte.com> Signed-off-by: Dominique Martinet <asmadeus@codewreck.org> 
Use a macro to define the default value for the 'msize' option
at one place instead of using two separate integer literals.

Link: http://lkml.kernel.org/r/28bb651ae0349a7d57e8ddc92c1bd5e62924a912.1630770829.git.linux_oss@crudebyte.com
Signed-off-by: Christian Schoenebeck <linux_oss@crudebyte.com>
Signed-off-by: Dominique Martinet <asmadeus@codewreck.org>
net/9p: increase tcp max msize to 1MB 2021-09-04T23:35:56+00:00 Dominique Martinet asmadeus@codewreck.org 2021-09-04T23:29:22+00:00 22bb3b79290ec5970b74fa6e9eb313802d075c82 Historically TCP has been limited to 64K buffers, but increasing msize provides huge performance benefits especially as latency increase so allow for bigger buffers. Ideally further improvements could change the allocation from the current contiguous chunk in slab (kmem_cache) to some scatter-gather compatible API... Note this only increases the max possible setting, not the default value. Link: http://lkml.kernel.org/r/YTQB5jCbvhmCWzNd@codewreck.org Signed-off-by: Dominique Martinet <asmadeus@codewreck.org> 
Historically TCP has been limited to 64K buffers, but increasing
msize provides huge performance benefits especially as latency
increase so allow for bigger buffers.

Ideally further improvements could change the allocation from the
current contiguous chunk in slab (kmem_cache) to some scatter-gather
compatible API...

Note this only increases the max possible setting, not the default
value.

Link: http://lkml.kernel.org/r/YTQB5jCbvhmCWzNd@codewreck.org
Signed-off-by: Dominique Martinet <asmadeus@codewreck.org>
9p/xen: Fix end of loop tests for list_for_each_entry 2021-07-27T14:01:16+00:00 Harshvardhan Jha harshvardhan.jha@oracle.com 2021-07-27T00:07:10+00:00 732b33d0dbf17e9483f0b50385bf606f724f50a2 This patch addresses the following problems: - priv can never be NULL, so this part of the check is useless - if the loop ran through the whole list, priv->client is invalid and it is more appropriate and sufficient to check for the end of list_for_each_entry loop condition. Link: http://lkml.kernel.org/r/20210727000709.225032-1-harshvardhan.jha@oracle.com Signed-off-by: Harshvardhan Jha <harshvardhan.jha@oracle.com> Reviewed-by: Stefano Stabellini <sstabellini@kernel.org> Tested-by: Stefano Stabellini <sstabellini@kernel.org> Cc: <stable@vger.kernel.org> Signed-off-by: Dominique Martinet <asmadeus@codewreck.org> 
This patch addresses the following problems:
 - priv can never be NULL, so this part of the check is useless
 - if the loop ran through the whole list, priv->client is invalid and
it is more appropriate and sufficient to check for the end of
list_for_each_entry loop condition.

Link: http://lkml.kernel.org/r/20210727000709.225032-1-harshvardhan.jha@oracle.com
Signed-off-by: Harshvardhan Jha <harshvardhan.jha@oracle.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
Tested-by: Stefano Stabellini <sstabellini@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Dominique Martinet <asmadeus@codewreck.org>
9p/trans_virtio: Remove sysfs file on probe failure 2021-07-27T14:01:15+00:00 Xie Yongji xieyongji@bytedance.com 2021-05-17T08:35:57+00:00 f997ea3b7afc108eb9761f321b57de2d089c7c48 This ensures we don't leak the sysfs file if we failed to allocate chan->vc_wq during probe. Link: http://lkml.kernel.org/r/20210517083557.172-1-xieyongji@bytedance.com Fixes: 86c8437383ac ("net/9p: Add sysfs mount_tag file for virtio 9P device") Signed-off-by: Xie Yongji <xieyongji@bytedance.com> Signed-off-by: Dominique Martinet <asmadeus@codewreck.org> 
This ensures we don't leak the sysfs file if we failed to
allocate chan->vc_wq during probe.

Link: http://lkml.kernel.org/r/20210517083557.172-1-xieyongji@bytedance.com
Fixes: 86c8437383ac ("net/9p: Add sysfs mount_tag file for virtio 9P device")
Signed-off-by: Xie Yongji <xieyongji@bytedance.com>
Signed-off-by: Dominique Martinet <asmadeus@codewreck.org>
9p/trans_virtio: Fix spelling mistakes 2021-06-02T21:01:55+00:00 Zheng Yongjun zhengyongjun3@huawei.com 2021-06-02T06:54:42+00:00 8ab1784df65176ad24b6d1e1376b8dd53ce2b695 reseting ==> resetting alloced ==> allocated accomodate ==> accommodate Signed-off-by: Zheng Yongjun <zhengyongjun3@huawei.com> Signed-off-by: David S. Miller <davem@davemloft.net> 
reseting  ==> resetting
alloced  ==> allocated
accomodate  ==> accommodate

Signed-off-by: Zheng Yongjun <zhengyongjun3@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net: 9p: Correct function names in the kerneldoc comments 2021-03-29T00:56:56+00:00 Xiongfeng Wang wangxiongfeng2@huawei.com 2021-03-27T08:15:54+00:00 03ff7371cba41903b9f53617a44093051ead3fe7 Fix the following W=1 kernel build warning(s): net/9p/client.c:133: warning: expecting prototype for parse_options(). Prototype was for parse_opts() instead net/9p/client.c:269: warning: expecting prototype for p9_req_alloc(). Prototype was for p9_tag_alloc() instead Reported-by: Hulk Robot <hulkci@huawei.com> Signed-off-by: Xiongfeng Wang <wangxiongfeng2@huawei.com> Signed-off-by: David S. Miller <davem@davemloft.net> 
Fix the following W=1 kernel build warning(s):

 net/9p/client.c:133: warning: expecting prototype for parse_options(). Prototype was for parse_opts() instead
 net/9p/client.c:269: warning: expecting prototype for p9_req_alloc(). Prototype was for p9_tag_alloc() instead

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Xiongfeng Wang <wangxiongfeng2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9p/trans_fd: Correct function name p9_mux_destroy() in the kerneldoc 2021-03-29T00:56:56+00:00 Xiongfeng Wang wangxiongfeng2@huawei.com 2021-03-27T08:15:53+00:00 54e625e3bd1dd59f6d0b95730fa9be2604aceb1c Fix the following W=1 kernel build warning(s): net/9p/trans_fd.c:881: warning: expecting prototype for p9_mux_destroy(). Prototype was for p9_conn_destroy() instead Reported-by: Hulk Robot <hulkci@huawei.com> Signed-off-by: Xiongfeng Wang <wangxiongfeng2@huawei.com> Signed-off-by: David S. Miller <davem@davemloft.net> 
Fix the following W=1 kernel build warning(s):

 net/9p/trans_fd.c:881: warning: expecting prototype for p9_mux_destroy(). Prototype was for p9_conn_destroy() instead

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Xiongfeng Wang <wangxiongfeng2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net: 9p: Correct function name errstr2errno() in the kerneldoc comments 2021-03-29T00:56:56+00:00 Xiongfeng Wang wangxiongfeng2@huawei.com 2021-03-27T08:15:52+00:00 8bf94a92505e6e9d46a76230b4900238685aa2ae Fix the following W=1 kernel build warning(s): net/9p/error.c:207: warning: expecting prototype for errstr2errno(). Prototype was for p9_errstr2errno() instead Reported-by: Hulk Robot <hulkci@huawei.com> Signed-off-by: Xiongfeng Wang <wangxiongfeng2@huawei.com> Signed-off-by: David S. Miller <davem@davemloft.net> 
Fix the following W=1 kernel build warning(s):

 net/9p/error.c:207: warning: expecting prototype for errstr2errno(). Prototype was for p9_errstr2errno() instead

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Xiongfeng Wang <wangxiongfeng2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net: 9p: advance iov on empty read 2021-03-04T00:57:59+00:00 Jisheng Zhang Jisheng.Zhang@synaptics.com 2021-03-02T09:19:32+00:00 d65614a01d24704b016635abf5cc028a54e45a62 I met below warning when cating a small size(about 80bytes) txt file on 9pfs(msize=2097152 is passed to 9p mount option), the reason is we miss iov_iter_advance() if the read count is 0 for zerocopy case, so we didn't truncate the pipe, then iov_iter_pipe() thinks the pipe is full. Fix it by removing the exception for 0 to ensure to call iov_iter_advance() even on empty read for zerocopy case. [ 8.279568] WARNING: CPU: 0 PID: 39 at lib/iov_iter.c:1203 iov_iter_pipe+0x31/0x40 [ 8.280028] Modules linked in: [ 8.280561] CPU: 0 PID: 39 Comm: cat Not tainted 5.11.0+ #6 [ 8.281260] RIP: 0010:iov_iter_pipe+0x31/0x40 [ 8.281974] Code: 2b 42 54 39 42 5c 76 22 c7 07 20 00 00 00 48 89 57 18 8b 42 50 48 c7 47 08 b [ 8.283169] RSP: 0018:ffff888000cbbd80 EFLAGS: 00000246 [ 8.283512] RAX: 0000000000000010 RBX: ffff888000117d00 RCX: 0000000000000000 [ 8.283876] RDX: ffff88800031d600 RSI: 0000000000000000 RDI: ffff888000cbbd90 [ 8.284244] RBP: ffff888000cbbe38 R08: 0000000000000000 R09: ffff8880008d2058 [ 8.284605] R10: 0000000000000002 R11: ffff888000375510 R12: 0000000000000050 [ 8.284964] R13: ffff888000cbbe80 R14: 0000000000000050 R15: ffff88800031d600 [ 8.285439] FS: 00007f24fd8af600(0000) GS:ffff88803ec00000(0000) knlGS:0000000000000000 [ 8.285844] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 8.286150] CR2: 00007f24fd7d7b90 CR3: 0000000000c97000 CR4: 00000000000406b0 [ 8.286710] Call Trace: [ 8.288279] generic_file_splice_read+0x31/0x1a0 [ 8.289273] ? do_splice_to+0x2f/0x90 [ 8.289511] splice_direct_to_actor+0xcc/0x220 [ 8.289788] ? pipe_to_sendpage+0xa0/0xa0 [ 8.290052] do_splice_direct+0x8b/0xd0 [ 8.290314] do_sendfile+0x1ad/0x470 [ 8.290576] do_syscall_64+0x2d/0x40 [ 8.290818] entry_SYSCALL_64_after_hwframe+0x44/0xa9 [ 8.291409] RIP: 0033:0x7f24fd7dca0a [ 8.292511] Code: c3 0f 1f 80 00 00 00 00 4c 89 d2 4c 89 c6 e9 bd fd ff ff 0f 1f 44 00 00 31 8 [ 8.293360] RSP: 002b:00007ffc20932818 EFLAGS: 00000206 ORIG_RAX: 0000000000000028 [ 8.293800] RAX: ffffffffffffffda RBX: 0000000001000000 RCX: 00007f24fd7dca0a [ 8.294153] RDX: 0000000000000000 RSI: 0000000000000003 RDI: 0000000000000001 [ 8.294504] RBP: 0000000000000003 R08: 0000000000000000 R09: 0000000000000000 [ 8.294867] R10: 0000000001000000 R11: 0000000000000206 R12: 0000000000000003 [ 8.295217] R13: 0000000000000001 R14: 0000000000000001 R15: 0000000000000000 [ 8.295782] ---[ end trace 63317af81b3ca24b ]--- Signed-off-by: Jisheng Zhang <Jisheng.Zhang@synaptics.com> Signed-off-by: David S. Miller <davem@davemloft.net> 
I met below warning when cating a small size(about 80bytes) txt file
on 9pfs(msize=2097152 is passed to 9p mount option), the reason is we
miss iov_iter_advance() if the read count is 0 for zerocopy case, so
we didn't truncate the pipe, then iov_iter_pipe() thinks the pipe is
full. Fix it by removing the exception for 0 to ensure to call
iov_iter_advance() even on empty read for zerocopy case.

[    8.279568] WARNING: CPU: 0 PID: 39 at lib/iov_iter.c:1203 iov_iter_pipe+0x31/0x40
[    8.280028] Modules linked in:
[    8.280561] CPU: 0 PID: 39 Comm: cat Not tainted 5.11.0+ #6
[    8.281260] RIP: 0010:iov_iter_pipe+0x31/0x40
[    8.281974] Code: 2b 42 54 39 42 5c 76 22 c7 07 20 00 00 00 48 89 57 18 8b 42 50 48 c7 47 08 b
[    8.283169] RSP: 0018:ffff888000cbbd80 EFLAGS: 00000246
[    8.283512] RAX: 0000000000000010 RBX: ffff888000117d00 RCX: 0000000000000000
[    8.283876] RDX: ffff88800031d600 RSI: 0000000000000000 RDI: ffff888000cbbd90
[    8.284244] RBP: ffff888000cbbe38 R08: 0000000000000000 R09: ffff8880008d2058
[    8.284605] R10: 0000000000000002 R11: ffff888000375510 R12: 0000000000000050
[    8.284964] R13: ffff888000cbbe80 R14: 0000000000000050 R15: ffff88800031d600
[    8.285439] FS:  00007f24fd8af600(0000) GS:ffff88803ec00000(0000) knlGS:0000000000000000
[    8.285844] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[    8.286150] CR2: 00007f24fd7d7b90 CR3: 0000000000c97000 CR4: 00000000000406b0
[    8.286710] Call Trace:
[    8.288279]  generic_file_splice_read+0x31/0x1a0
[    8.289273]  ? do_splice_to+0x2f/0x90
[    8.289511]  splice_direct_to_actor+0xcc/0x220
[    8.289788]  ? pipe_to_sendpage+0xa0/0xa0
[    8.290052]  do_splice_direct+0x8b/0xd0
[    8.290314]  do_sendfile+0x1ad/0x470
[    8.290576]  do_syscall_64+0x2d/0x40
[    8.290818]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
[    8.291409] RIP: 0033:0x7f24fd7dca0a
[    8.292511] Code: c3 0f 1f 80 00 00 00 00 4c 89 d2 4c 89 c6 e9 bd fd ff ff 0f 1f 44 00 00 31 8
[    8.293360] RSP: 002b:00007ffc20932818 EFLAGS: 00000206 ORIG_RAX: 0000000000000028
[    8.293800] RAX: ffffffffffffffda RBX: 0000000001000000 RCX: 00007f24fd7dca0a
[    8.294153] RDX: 0000000000000000 RSI: 0000000000000003 RDI: 0000000000000001
[    8.294504] RBP: 0000000000000003 R08: 0000000000000000 R09: 0000000000000000
[    8.294867] R10: 0000000001000000 R11: 0000000000000206 R12: 0000000000000003
[    8.295217] R13: 0000000000000001 R14: 0000000000000001 R15: 0000000000000000
[    8.295782] ---[ end trace 63317af81b3ca24b ]---

Signed-off-by: Jisheng Zhang <Jisheng.Zhang@synaptics.com>
Signed-off-by: David S. Miller <davem@davemloft.net>