On Wed 04-09-24 12:15:15, Kent Overstreet wrote: > On Tue, Sep 03, 2024 at 09:06:17AM GMT, Michal Hocko wrote: > > On Mon 02-09-24 18:32:33, Kent Overstreet wrote: [...] > > > For bcachefs: I try really hard to minimize tail latency and make > > > performance robust in extreme scenarios - thrashing. A large part of > > > that is that btree locks must be held for no longer than necessary. > > > > > > We definitely don't want to recurse into other parts of the kernel, > > > taking other locks (i.e. in memory reclaim) while holding btree locks; > > > that's a great way to stack up (and potentially multiply) latencies. > > > > OK, these two patches do not fail to do that. The only existing user is > > turned into GFP_NOWAIT so the final code works the same way. Right? > > https://lore.kernel.org/linux-mm/20240828140638.3204253-1-kent.overstreet@xxxxxxxxx/ https://lore.kernel.org/linux-mm/Zs9xC3OJPbkMy25C@xxxxxxxxxxxxxxxxxxxx/ > > > But gfp flags don't work with vmalloc allocations (and that's unlikely > > > to change), and we require vmalloc fallbacks for e.g. btree node > > > allocation. That's the big reason we want MEMALLOC_PF_NORECLAIM. > > > > Have you even tried to reach out to vmalloc maintainers and asked for > > GFP_NOWAIT support for vmalloc? Because I do not remember that. Sure > > kernel page tables are have hardcoded GFP_KERNEL context which slightly > > complicates that but that doesn't really mean the only potential > > solution is to use a per task flag to override that. Just from top of my > > head we can consider pre-allocating virtual address space for > > non-sleeping allocations. Maybe there are other options that only people > > deeply familiar with the vmalloc internals can see. > > That sounds really overly complicated. Let vmalloc people discuss viable ways to deal with that. You as vmalloc consumer want to get NOWAIT support. Ask them and see what kind of solution they can offer to you as a user. This is how we develop kernel in a collaborative way. We do not enforce solutions we work with domain experts to work out a maintainable solution. -- Michal Hocko SUSE Labs
