On Wed, 24 Jun 2020, Srikar Dronamraju wrote:
> Currently Linux kernel with CONFIG_NUMA on a system with multiple
> possible nodes, marks node 0 as online at boot. However in practice,
> there are systems which have node 0 as memoryless and cpuless.
Maybe add something to explain why you are not
On Tue, 12 May 2020, Srikar Dronamraju wrote:
> +#ifdef CONFIG_NUMA
> + [N_ONLINE] = NODE_MASK_NONE,
Again. Same issue as before. If you do this then you do a global change
for all architectures. You need to put something in the early boot
sequence (in a non architecture specific way) that se
On Fri, 1 May 2020, Srikar Dronamraju wrote:
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -116,8 +116,10 @@ EXPORT_SYMBOL(latent_entropy);
> */
> nodemask_t node_states[NR_NODE_STATES] __read_mostly = {
> [N_POSSIBLE] = NODE_MASK_ALL,
> +#ifdef CONFIG_NUMA
> + [N_ONLINE] = NOD
On Fri, 1 May 2020, Srikar Dronamraju wrote:
> - for_each_present_cpu(cpu)
> - numa_setup_cpu(cpu);
> + for_each_possible_cpu(cpu) {
> + /*
> + * Powerpc with CONFIG_NUMA always used to have a node 0,
> + * even if it was memoryless or cpul
On Wed, 18 Mar 2020, Srikar Dronamraju wrote:
> For a memoryless or offline nodes, node_numa_mem refers to a N_MEMORY
> fallback node. Currently kernel has an API set_numa_mem that sets
> node_numa_mem for memoryless node. However this API cannot be used for
> offline nodes. Hence all offline node
On Mon, 16 Mar 2020, Michal Hocko wrote:
> > We can dynamically number the nodes right? So just make sure that the
> > firmware properly creates memory on node 0?
>
> Are you suggesting that the OS would renumber NUMA nodes coming
> from FW just to satisfy node 0 existence? If yes then I believe t
On Wed, 11 Mar 2020, Srikar Dronamraju wrote:
> Currently Linux kernel with CONFIG_NUMA on a system with multiple
> possible nodes, marks node 0 as online at boot. However in practice,
> there are systems which have node 0 as memoryless and cpuless.
Would it not be better and simpler to require
On Wed, 26 Feb 2020, Michal Hocko wrote:
> Besides that kmalloc_node shouldn't really have an implicit GFP_THISNODE
> semantic right? At least I do not see anything like that documented
> anywhere.
Kmalloc_node does not support memory policies etc. Only kmalloc does.
kmalloc_node is mostly used b
On Mon, 24 Feb 2020, Michal Hocko wrote:
> Hmm, nasty. Is there any reason why kmalloc_node behaves differently
> from the page allocator?
The page allocator will do the same thing if you pass GFP_THISNODE and
insist on allocating memory from a node that does not exist.
> > > A short summary. k
On Tue, 18 Feb 2020, Michal Hocko wrote:
> Anyway, I do not think it is expected that kmalloc_node just blows up
> on those nodes. The page allocator simply falls back to the closest
> node. Something for kmalloc maintainers I believe.
That is the case for an unconstrained allocation. kmalloc_nod
On Tue, 8 Oct 2019, Leonardo Bras wrote:
> So you say that the performance impact of using my approach is the same
> as using locks? (supposing that lock never waits)
>
> So, there are 'lockless pagetable walks' only for the sake of better
> performance?
I thought that was the major motivation
On Tue, 8 Oct 2019, Leonardo Bras wrote:
> > You are creating contention on a single exclusive cacheline. Doesnt this
> > defeat the whole purpose of the lockless page table walk? Use mmap_sem or
> > so should cause the same performance regression?
>
> Sorry, I did not understand that question.
>
On Wed, 2 Oct 2019, Leonardo Bras wrote:
> +
> +inline unsigned long __begin_lockless_pgtbl_walk(struct mm_struct *mm,
> + bool disable_irq)
> +{
> + unsigned long irq_mask = 0;
> +
> + if (IS_ENABLED(CONFIG_LOCKLESS_PAGE_TABLE_WALK_TRACKING)
On Thu, 14 Feb 2019, Jason Gunthorpe wrote:
> On Thu, Feb 14, 2019 at 01:46:51PM -0800, Ira Weiny wrote:
>
> > > > > Really unclear how to fix this. The pinned/locked split with two
> > > > > buckets may be the right way.
> > > >
> > > > Are you suggesting that we have 2 user limits?
> > >
> > > T
On Tue, 12 Feb 2019, Alexey Kardashevskiy wrote:
> Now it is 3 independent accesses (actually 4 but the last one is
> diagnostic) with no locking around them. Why do not we need a lock
> anymore precisely? Thanks,
Updating a regular counter is racy and requires a lock. It was converted
to be an a
15 matches
Mail list logo
