Michael S. Tsirkin wrote: > > Yes, conditional GFP_KERNEL allocation attempt from virtqueue_add() might > > still cause this deadlock. But that depends on whether you can trigger this > > deadlock. As far as I know, there is no report. Thus, I think that avoiding > > theoretical deadlock using timeout will be sufficient. > > > So first of all IMHO GFP_KERNEL allocations do not happen in > virtqueue_add_outbuf at all. They only trigger through add_sgs. I did not notice that total_sg == 1 is true for virtqueue_add_outbuf(). > > IMHO this is an API bug, we should just drop the gfp parameter > from this API. OK. > > > so the issue is balloon_page_enqueue only. > Since you explained that there is "the deflate on OOM flag", we don't want to skip deflating upon lock contention. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>
