Messages in this thread | | | Date | Tue, 1 Sep 2015 13:39:39 +0200 | From | Peter Zijlstra <> | Subject | Re: [PATCH v2][RESEND] perf, x86: Fix multi-segment problem of perf_event_intel_uncore |
| |
On Tue, Sep 01, 2015 at 01:31:47PM +0200, Ingo Molnar wrote: > > * Peter Zijlstra <peterz@infradead.org> wrote: > > > +struct __find_pci2phy(int segment) > > +{ > > + struct pci2phy_map *map; > > + > > + lockdep_assert_held(&pci2phy_map_lock); > > + > > + list_for_each_entry(map, &pci2phy_map_head, list) { > > + if (map->segment == segment) > > + return map; > > + } > > + > > + map = kmalloc(sizeof(struct pci2phy_map), GFP_ATOMIC); > > I really don't think we should be adding new GFP_ATOMIC allocations whenever it > can be avoided.
I tend to agree, it makes the whole thing rather fragile...
| |