Re: [PATCH v2] sock: allow reading and changing sk_userlocks with setsockopt

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 





On 02.08.2021 19:11, Jakub Kicinski wrote:
On Mon, 2 Aug 2021 11:26:09 +0300 Pavel Tikhomirov wrote:
On 30.07.2021 19:46, Jakub Kicinski wrote:
On Fri, 30 Jul 2021 19:07:08 +0300 Pavel Tikhomirov wrote:
SOCK_SNDBUF_LOCK and SOCK_RCVBUF_LOCK flags disable automatic socket
buffers adjustment done by kernel (see tcp_fixup_rcvbuf() and
tcp_sndbuf_expand()). If we've just created a new socket this adjustment
is enabled on it, but if one changes the socket buffer size by
setsockopt(SO_{SND,RCV}BUF*) it becomes disabled.

CRIU needs to call setsockopt(SO_{SND,RCV}BUF*) on each socket on
restore as it first needs to increase buffer sizes for packet queues
restore and second it needs to restore back original buffer sizes. So
after CRIU restore all sockets become non-auto-adjustable, which can
decrease network performance of restored applications significantly.

CRIU need to be able to restore sockets with enabled/disabled adjustment
to the same state it was before dump, so let's add special setsockopt
for it.

Signed-off-by: Pavel Tikhomirov <ptikhomirov@xxxxxxxxxxxxx>

The patchwork bot is struggling to ingest this, please double check it
applies cleanly to net-next.

I checked that it applies cleanly to net-next:

[snorch@fedora linux]$ git am
~/Downloads/patches/ptikhomirov/setsockopt-sk_userlocks/\[PATCH\ v2\]\
sock\:\ allow\ reading\ and\ changing\ sk_userlocks\ with\ setsockopt.eml

[snorch@fedora linux]$ git log --oneline
c339520aadd5 (HEAD -> net-next) sock: allow reading and changing
sk_userlocks with setsockopt

d39e8b92c341 (net-next/master) Merge
https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next

Probably it was some temporary problem and now it's OK?
https://patchwork.kernel.org/project/netdevbpf/patch/20210730160708.6544-1-ptikhomirov@xxxxxxxxxxxxx/

Indeed, must have been resolved by the merge of net into net-next which
happened on Saturday? Regardless, would you mind reposting? There is no
way for me to retry the patchwork checks.

And one more thing..

+	case SO_BUF_LOCK:
+		sk->sk_userlocks = (sk->sk_userlocks & ~SOCK_BUF_LOCK_MASK) |
+				   (val & SOCK_BUF_LOCK_MASK);

What's the thinking on silently ignoring unsupported flags on set
rather than rejecting? I feel like these days we lean towards explicit
rejects.

Will do.


+	case SO_BUF_LOCK:
+		v.val = sk->sk_userlocks & (SOCK_SNDBUF_LOCK | SOCK_RCVBUF_LOCK);
+		break;

The mask could you be used here.

Sure, missed it...


Just to double check - is the expectation that the value returned is
completely opaque to the user space? The defines in question are not
part of uAPI.

Sorry, didn't though about it initially. For criu we don't care about the actual bits we restore same what we've dumped. Buf if some real users would like to use this interface to restore default autoadjustment on their sockets we should probably export SOCK_SNDBUF_LOCK and SOCK_RCVBUF_LOCK to uAPI.



--
Best regards, Tikhomirov Pavel
Software Developer, Virtuozzo.



[Index of Archives]     [Netdev]     [Linux Wireless]     [Kernel Newbies]     [Security]     [Linux for Hams]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux RAID]     [Linux Admin]     [Samba]

  Powered by Linux