On Mon, 2018-04-09 at 11:17 -0600, Jason Gunthorpe wrote: > On Sun, Apr 08, 2018 at 01:08:57AM +0000, Bart Van Assche wrote: > > On Sun, 2018-04-08 at 08:54 +0800, Yanjun Zhu wrote: > > > > > > On 2018/3/31 12:43, Bart Van Assche wrote: > > > > On Sat, 2018-03-31 at 00:18 -0400, Zhu Yanjun wrote: > > > > > + refcount_inc(&res->atomic.skb->users); > > > > > > > > Is this perhaps an open-coded skb_get() call? If so, why isn't skb_get() > > > > called here? > > > > > > Thanks. There are some places that use refcount_inc instead of skb_get. > > > I will fix it with skb_get later in a patch. > > > > That's a weird approach. What prevents you from modifying this patch such > > that it uses skb_get()? > > Bart, did the regression you reported in rxe get fixed? I don't really > want to apply more patches until I see a fix.. Hello Jason, I'm currently busy with working on a fix for a block layer regression. I will retest the rdma_rxe driver as soon as I have finished with this. See also "[PATCH] blk-mq: Fix recently introduced races in the timeout handling code" (https://www.mail-archive.com/linux-block@xxxxxxxxxxxxxxx/msg20073.html). Thanks, Bart. ��.n��������+%������w��{.n�����{���fk��ܨ}���Ơz�j:+v�����w����ޙ��&�)ߡ�a����z�ޗ���ݢj��w�f
