hugetlb parallelization depends on PADATA, and PADATA depends on SMP.
PADATA consists of two distinct functionality: One part is
padata_do_multithreaded which disregards order and simply divides tasks
into several groups for parallel execution. Hugetlb init parallelization
depends on padata_do_multithreaded.
The other part is composed of a set of APIs that, while handling data in
an out-of-order parallel manner, can eventually return the data with
ordered sequence. Currently Only `crypto/pcrypt.c` use them.
All users of PADATA of non-SMP case currently only use
padata_do_multithreaded. It is easy to implement a serial one in
include/linux/padata.h. And it is not necessary to implement another
functionality unless the only user of crypto/pcrypt.c does not depend on
SMP in the future.
Link: https://lkml.kernel.org/r/20240222140422.393911-6-gang.li@linux.dev
Signed-off-by: Gang Li <ligang.bdlg@bytedance.com>
Tested-by: Paul E. McKenney <paulmck@kernel.org>
Acked-by: Daniel Jordan <daniel.m.jordan@oracle.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Jane Chu <jane.chu@oracle.com>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Steffen Klassert <steffen.klassert@secunet.com>
Cc: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "hugetlb: parallelize hugetlb page init on boot", v6.
Introduction
------------
Hugetlb initialization during boot takes up a considerable amount of time.
For instance, on a 2TB system, initializing 1,800 1GB huge pages takes
1-2 seconds out of 10 seconds. Initializing 11,776 1GB pages on a 12TB
Intel host takes more than 1 minute[1]. This is a noteworthy figure.
Inspired by [2] and [3], hugetlb initialization can also be accelerated
through parallelization. Kernel already has infrastructure like
padata_do_multithreaded, this patch uses it to achieve effective results
by minimal modifications.
[1] https://lore.kernel.org/all/783f8bac-55b8-5b95-eb6a-11a583675000@google.com/
[2] https://lore.kernel.org/all/20200527173608.2885243-1-daniel.m.jordan@oracle.com/
[3] https://lore.kernel.org/all/20230906112605.2286994-1-usama.arif@bytedance.com/
[4] https://lore.kernel.org/all/76becfc1-e609-e3e8-2966-4053143170b6@google.com/
max_threads
-----------
This patch use `padata_do_multithreaded` like this:
```
job.max_threads = num_node_state(N_MEMORY) * multiplier;
padata_do_multithreaded(&job);
```
To fully utilize the CPU, the number of parallel threads needs to be
carefully considered. `max_threads = num_node_state(N_MEMORY)` does not
fully utilize the CPU, so we need to multiply it by a multiplier.
Tests below indicate that a multiplier of 2 significantly improves
performance, and although larger values also provide improvements, the
gains are marginal.
multiplier 1 2 3 4 5
------------ ------- ------- ------- ------- -------
256G 2node 358ms 215ms 157ms 134ms 126ms
2T 4node 979ms 679ms 543ms 489ms 481ms
50G 2node 71ms 44ms 37ms 30ms 31ms
Therefore, choosing 2 as the multiplier strikes a good balance between
enhancing parallel processing capabilities and maintaining efficient
resource management.
Test result
-----------
test case no patch(ms) patched(ms) saved
------------------- -------------- ------------- --------
256c2T(4 node) 1G 4745 2024 57.34%
128c1T(2 node) 1G 3358 1712 49.02%
12T 1G 77000 18300 76.23%
256c2T(4 node) 2M 3336 1051 68.52%
128c1T(2 node) 2M 1943 716 63.15%
This patch (of 8):
The readability of `hugetlb_hstate_alloc_pages` is poor. By cleaning the
code, its readability can be improved, facilitating future modifications.
This patch extracts two functions to reduce the complexity of
`hugetlb_hstate_alloc_pages` and has no functional changes.
- hugetlb_hstate_alloc_pages_node_specific() to handle iterates through
each online node and performs allocation if necessary.
- hugetlb_hstate_alloc_pages_report() report error during allocation.
And the value of h->max_huge_pages is updated accordingly.
Link: https://lkml.kernel.org/r/20240222140422.393911-1-gang.li@linux.dev
Link: https://lkml.kernel.org/r/20240222140422.393911-2-gang.li@linux.dev
Signed-off-by: Gang Li <ligang.bdlg@bytedance.com>
Tested-by: David Rientjes <rientjes@google.com>
Reviewed-by: Muchun Song <muchun.song@linux.dev>
Reviewed-by: Tim Chen <tim.c.chen@linux.intel.com>
Cc: Daniel Jordan <daniel.m.jordan@oracle.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jane Chu <jane.chu@oracle.com>
Cc: Paul E. McKenney <paulmck@kernel.org>
Cc: Randy Dunlap <rdunlap@infradead.org>
Cc: Steffen Klassert <steffen.klassert@secunet.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
do_numa_page() is reading from the same page table entry, twice, while
holding the page table lock: once while checking that the pte hasn't
changed, and again in order to modify the pte.
Instead, just read the pte once, and save it in the same old_pte variable
that already exists. This has no effect on behavior, other than to
provide a tiny potential improvement to performance, by avoiding the
redundant memory read (which the compiler cannot elide, due to
READ_ONCE()).
Also improve the associated comments nearby.
Link: https://lkml.kernel.org/r/20240228034151.459370-1-jhubbard@nvidia.com
Signed-off-by: John Hubbard <jhubbard@nvidia.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
The synchronization here is to ensure the ordering of freeing of a module
init so that it happens before W+X checking. It is worth noting it is not
that the freeing was not happening, it is just that our sanity checkers
raced against the permission checkers which assume init memory is already
gone.
Commit 1a7b7d9220 ("modules: Use vmalloc special flag") moved calling
do_free_init() into a global workqueue instead of relying on it being
called through call_rcu(..., do_free_init), which used to allowed us call
do_free_init() asynchronously after the end of a subsequent grace period.
The move to a global workqueue broke the gaurantees for code which needed
to be sure the do_free_init() would complete with rcu_barrier(). To fix
this callers which used to rely on rcu_barrier() must now instead use
flush_work(&init_free_wq).
Without this fix, we still could encounter false positive reports in W+X
checking since the rcu_barrier() here can not ensure the ordering now.
Even worse, the rcu_barrier() can introduce significant delay. Eric
Chanudet reported that the rcu_barrier introduces ~0.1s delay on a
PREEMPT_RT kernel.
[ 0.291444] Freeing unused kernel memory: 5568K
[ 0.402442] Run /sbin/init as init process
With this fix, the above delay can be eliminated.
Link: https://lkml.kernel.org/r/20240227023546.2490667-1-changbin.du@huawei.com
Fixes: 1a7b7d9220 ("modules: Use vmalloc special flag")
Signed-off-by: Changbin Du <changbin.du@huawei.com>
Tested-by: Eric Chanudet <echanude@redhat.com>
Acked-by: Luis Chamberlain <mcgrof@kernel.org>
Cc: Xiaoyi Su <suxiaoyi@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "Rearrange batched folio freeing", v3.
Other than the obvious "remove calls to compound_head" changes, the
fundamental belief here is that iterating a linked list is much slower
than iterating an array (5-15x slower in my testing). There's also an
associated belief that since we iterate the batch of folios three times,
we do better when the array is small (ie 15 entries) than we do with a
batch that is hundreds of entries long, which only gives us the
opportunity for the first pages to fall out of cache by the time we get to
the end.
It is possible we should increase the size of folio_batch. Hopefully the
bots let us know if this introduces any performance regressions.
This patch (of 3):
By making release_pages() call folios_put(), we can get rid of the calls
to compound_head() for the callers that already know they have folios. We
can also get rid of the lock_batch tracking as we know the size of the
batch is limited by folio_batch. This does reduce the maximum number of
pages for which the lruvec lock is held, from SWAP_CLUSTER_MAX (32) to
PAGEVEC_SIZE (15). I do not expect this to make a significant difference,
but if it does, we can increase PAGEVEC_SIZE to 31.
Link: https://lkml.kernel.org/r/20240227174254.710559-1-willy@infradead.org
Link: https://lkml.kernel.org/r/20240227174254.710559-2-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Patch series "mm: remove total_mapcount()", v2.
Let's remove the remaining user from mm/memfd.c so we can get rid of
total_mapcount().
This patch (of 2):
Both functions are the remaining users of total_mapcount(). Let's get rid
of the calls by converting the code to folios.
As it turns out, the code is unnecessarily complicated, especially:
1) We can query the number of pagecache references for a folio simply via
folio_nr_pages(). This will handle other folio sizes in the future
correctly.
2) The xas_set(xas, page->index + cache_count) call to increment the
iterator for large folios is not required. Remove it.
Further, simplify the XA_CHECK_SCHED check, counting each entry exactly
once.
Memfd pages can be swapped out when using shmem; leave xa_is_value()
checks in place.
Link: https://lkml.kernel.org/r/20240226141324.278526-1-david@redhat.com
Link: https://lkml.kernel.org/r/20240226141324.278526-2-david@redhat.com
Co-developed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
While doing MADV_PAGEOUT, the current code will clear PTE young so that
vmscan won't read young flags to allow the reclamation of madvised folios
to go ahead. It seems we can do it by directly ignoring references, thus
we can remove tlb flush in madvise and rmap overhead in vmscan.
Regarding the side effect, in the original code, if a parallel thread runs
side by side to access the madvised memory with the thread doing madvise,
folios will get a chance to be re-activated by vmscan (though the time gap
is actually quite small since checking PTEs is done immediately after
clearing PTEs young). But with this patch, they will still be reclaimed.
But this behaviour doing PAGEOUT and doing access at the same time is
quite silly like DoS. So probably, we don't need to care. Or ignoring
the new access during the quite small time gap is even better.
For DAMON's DAMOS_PAGEOUT based on physical address region, we still keep
its behaviour as is since a physical address might be mapped by multiple
processes. MADV_PAGEOUT based on virtual address is actually much more
aggressive on reclamation. To untouch paddr's DAMOS_PAGEOUT, we simply
pass ignore_references as false in reclaim_pages().
A microbench as below has shown 6% decrement on the latency of
MADV_PAGEOUT,
#define PGSIZE 4096
main()
{
int i;
#define SIZE 512*1024*1024
volatile long *p = mmap(NULL, SIZE, PROT_READ | PROT_WRITE,
MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
for (i = 0; i < SIZE/sizeof(long); i += PGSIZE / sizeof(long))
p[i] = 0x11;
madvise(p, SIZE, MADV_PAGEOUT);
}
w/o patch w/ patch
root@10:~# time ./a.out root@10:~# time ./a.out
real 0m49.634s real 0m46.334s
user 0m0.637s user 0m0.648s
sys 0m47.434s sys 0m44.265s
Link: https://lkml.kernel.org/r/20240226005739.24350-1-21cnbao@gmail.com
Signed-off-by: Barry Song <v-songbaohua@oppo.com>
Acked-by: Minchan Kim <minchan@kernel.org>
Cc: SeongJae Park <sj@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
The doc needs a fix. As only in the case of virtual address, we are
calling madvise() with MADV_PAGEOUT. But in the case of physical address,
we are calling reclaim_pages() directly. MADV_PAGEOUT on virtual address
is much more aggresive to reclaim memory compared to reclaim_pages() on
paddr region. This patch removes the details so that the description can
apply to both cases. And we don't need to couple with the implementation
details.
Link: https://lkml.kernel.org/r/20240224224751.4673-1-21cnbao@gmail.com
Signed-off-by: Barry Song <v-songbaohua@oppo.com>
Reviewed-by: SeongJae Park <sj@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
