write in 30M

hc

2024-02-20 ea08eeccae9297f7aabd2ef7f0c2517ac4549acc

 kernel/Documentation/admin-guide/mm/transhuge.rst
..
..
@@ -53,7 +53,7 @@
53
53
 collapses sequences of basic pages into huge pages.
54
54
 
55
55
 The THP behaviour is controlled via :ref:`sysfs <thp_sysfs>`
56
-interface and using madivse(2) and prctl(2) system calls.
56
+interface and using madvise(2) and prctl(2) system calls.
57
57
 
58
58
 Transparent Hugepage Support maximizes the usefulness of free memory
59
59
 if compared to the reservation approach of hugetlbfs by allowing all
..
..
@@ -220,6 +220,13 @@
220
220
 collapsed, resulting fewer pages being collapsed into
221
221
 THPs, and lower memory access performance.
222
222
 
223
+``max_ptes_shared`` specifies how many pages can be shared across multiple
224
+processes. Exceeding the number would block the collapse::
225
+
226
+	/sys/kernel/mm/transparent_hugepage/khugepaged/max_ptes_shared
227
+
228
+A higher value may increase memory footprint for some workloads.
229
+
223
230
 Boot parameter
224
231
 ==============
225
232
 
..
..
@@ -298,8 +305,7 @@
298
305
 
299
306
 thp_fault_alloc
300
307
 	is incremented every time a huge page is successfully
301
-	allocated to handle a page fault. This applies to both the
302
-	first time a page is faulted and for COW faults.
308
+	allocated to handle a page fault.
303
309
 
304
310
 thp_collapse_alloc
305
311
 	is incremented by khugepaged when it has found
..
..
@@ -310,6 +316,11 @@
310
316
 	is incremented if a page fault fails to allocate
311
317
 	a huge page and instead falls back to using small pages.
312
318
 
319
+thp_fault_fallback_charge
320
+	is incremented if a page fault fails to charge a huge page and
321
+	instead falls back to using small pages even though the
322
+	allocation was successful.
323
+
313
324
 thp_collapse_alloc_failed
314
325
 	is incremented if khugepaged found a range
315
326
 	of pages that should be collapsed into one huge page but failed
..
..
@@ -319,6 +330,15 @@
319
330
 	is incremented every time a file huge page is successfully
320
331
 	allocated.
321
332
 
333
+thp_file_fallback
334
+	is incremented if a file huge page is attempted to be allocated
335
+	but fails and instead falls back to using small pages.
336
+
337
+thp_file_fallback_charge
338
+	is incremented if a file huge page cannot be charged and instead
339
+	falls back to using small pages even though the allocation was
340
+	successful.
341
+
322
342
 thp_file_mapped
323
343
 	is incremented every time a file huge page is mapped into
324
344
 	user address space.