Remove dead pai_t field and delete pai.h

After replacing PAI vtable dispatch with direct calls in the previous
commit, the embedded pai_t member in pac_t and hpa_shard_t is dead
weight, and pai.h has no remaining users. Remove them.

Changes:
- Drop pai_t pai member (and "must be first member" comment) from
  pac_t and hpa_shard_t.
- Replace #include "jemalloc/internal/pai.h" with the actually-needed
  edata.h / tsd_types.h in pac.h, hpa.h, sec.h, pa.h.
- Update extent_pai_t comment in edata.h to no longer reference pai.h.
- Update three remaining test files (hpa_thp_always,
  hpa_vectorized_madvise, hpa_vectorized_madvise_large_batch) to call
  hpa_*(tsdn, shard, ...) directly instead of pai_*(tsdn, &shard->pai,
  ...).
- Delete include/jemalloc/internal/pai.h.

No behavioral changes.

include/jemalloc/internal/edata.h

@@ -48,8 +48,8 @@ enum extent_head_state_e {
 typedef enum extent_head_state_e extent_head_state_t;
 
 /*
- * Which implementation of the page allocator interface, (PAI, defined in
+ * Which page allocator implementation (PAC or HPA) owns the given extent?
- * pai.h) owns the given extent?
+ * Used by PA to route expand/shrink/dalloc to the correct implementation.
  */
 enum extent_pai_e { EXTENT_PAI_PAC = 0, EXTENT_PAI_HPA = 1 };
 typedef enum extent_pai_e extent_pai_t;

include/jemalloc/internal/hpa.h

@@ -10,7 +10,6 @@
 #include "jemalloc/internal/hpa_hooks.h"
 #include "jemalloc/internal/hpa_opts.h"
 #include "jemalloc/internal/mutex.h"
-#include "jemalloc/internal/pai.h"
 #include "jemalloc/internal/psset.h"
 #include "jemalloc/internal/sec.h"
 
@@ -63,12 +62,6 @@ struct hpa_shard_stats_s {
 
 typedef struct hpa_shard_s hpa_shard_t;
 struct hpa_shard_s {
-	/*
-	 * pai must be the first member; we cast from a pointer to it to a
-	 * pointer to the hpa_shard_t.
-	 */
-	pai_t pai;
-
 	/* The central allocator we get our hugepages from. */
 	hpa_central_t *central;
 

include/jemalloc/internal/pa.h

@@ -10,15 +10,13 @@
 #include "jemalloc/internal/hpa.h"
 #include "jemalloc/internal/lockedint.h"
 #include "jemalloc/internal/pac.h"
-#include "jemalloc/internal/pai.h"
 #include "jemalloc/internal/sec.h"
 
 /*
  * The page allocator; responsible for acquiring pages of memory for
- * allocations.  It picks the implementation of the page allocator interface
+ * allocations.  It dispatches each page-level allocation request to either
- * (i.e. a pai_t) to handle a given page-level allocation request.  For now, the
+ * the PAC ("page allocator classic") or the HPA ("hugepage allocator")
- * only such implementation is the PAC code ("page allocator classic"), but
+ * by calling their pac_*() / hpa_*() entry points directly.
- * others will be coming soon.
  */
 
 typedef struct pa_central_s pa_central_t;

include/jemalloc/internal/pac.h

@@ -4,10 +4,11 @@
 #include "jemalloc/internal/jemalloc_preamble.h"
 #include "jemalloc/internal/decay.h"
 #include "jemalloc/internal/ecache.h"
+#include "jemalloc/internal/edata.h"
 #include "jemalloc/internal/edata_cache.h"
 #include "jemalloc/internal/exp_grow.h"
 #include "jemalloc/internal/lockedint.h"
-#include "jemalloc/internal/pai.h"
+#include "jemalloc/internal/tsd_types.h"
 #include "san_bump.h"
 
 /*
@@ -87,11 +88,6 @@ struct pac_stats_s {
 
 typedef struct pac_s pac_t;
 struct pac_s {
-	/*
-	 * Must be the first member (we convert it to a PAC given only a
-	 * pointer).  The handle to the allocation interface.
-	 */
-	pai_t pai;
 	/* True once pinned memory has been seen. */
 	atomic_b_t has_pinned;
 	/*

include/jemalloc/internal/pai.h

@@ -1,63 +0,0 @@
-#ifndef JEMALLOC_INTERNAL_PAI_H
-#define JEMALLOC_INTERNAL_PAI_H
-
-#include "jemalloc/internal/jemalloc_preamble.h"
-#include "jemalloc/internal/edata.h"
-#include "jemalloc/internal/tsd_types.h"
-
-/* An interface for page allocation. */
-
-typedef struct pai_s pai_t;
-struct pai_s {
-	/* Returns NULL on failure. */
-	edata_t *(*alloc)(tsdn_t *tsdn, pai_t *self, size_t size,
-	    size_t alignment, bool zero, bool guarded, bool frequent_reuse,
-	    bool *deferred_work_generated);
-	bool (*expand)(tsdn_t *tsdn, pai_t *self, edata_t *edata,
-	    size_t old_size, size_t new_size, bool zero,
-	    bool *deferred_work_generated);
-	bool (*shrink)(tsdn_t *tsdn, pai_t *self, edata_t *edata,
-	    size_t old_size, size_t new_size, bool *deferred_work_generated);
-	void (*dalloc)(tsdn_t *tsdn, pai_t *self, edata_t *edata,
-	    bool *deferred_work_generated);
-	uint64_t (*time_until_deferred_work)(tsdn_t *tsdn, pai_t *self);
-};
-
-/*
- * These are just simple convenience functions to avoid having to reference the
- * same pai_t twice on every invocation.
- */
-
-static inline edata_t *
-pai_alloc(tsdn_t *tsdn, pai_t *self, size_t size, size_t alignment, bool zero,
-    bool guarded, bool frequent_reuse, bool *deferred_work_generated) {
-	return self->alloc(tsdn, self, size, alignment, zero, guarded,
-	    frequent_reuse, deferred_work_generated);
-}
-
-static inline bool
-pai_expand(tsdn_t *tsdn, pai_t *self, edata_t *edata, size_t old_size,
-    size_t new_size, bool zero, bool *deferred_work_generated) {
-	return self->expand(tsdn, self, edata, old_size, new_size, zero,
-	    deferred_work_generated);
-}
-
-static inline bool
-pai_shrink(tsdn_t *tsdn, pai_t *self, edata_t *edata, size_t old_size,
-    size_t new_size, bool *deferred_work_generated) {
-	return self->shrink(
-	    tsdn, self, edata, old_size, new_size, deferred_work_generated);
-}
-
-static inline void
-pai_dalloc(
-    tsdn_t *tsdn, pai_t *self, edata_t *edata, bool *deferred_work_generated) {
-	self->dalloc(tsdn, self, edata, deferred_work_generated);
-}
-
-static inline uint64_t
-pai_time_until_deferred_work(tsdn_t *tsdn, pai_t *self) {
-	return self->time_until_deferred_work(tsdn, self);
-}
-
-#endif /* JEMALLOC_INTERNAL_PAI_H */

include/jemalloc/internal/sec.h

@@ -4,9 +4,10 @@
 #include "jemalloc/internal/jemalloc_preamble.h"
 #include "jemalloc/internal/base.h"
 #include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/edata.h"
 #include "jemalloc/internal/mutex.h"
-#include "jemalloc/internal/pai.h"
 #include "jemalloc/internal/sec_opts.h"
+#include "jemalloc/internal/tsd_types.h"
 
 /*
  * Small extent cache.

test/unit/hpa_vectorized_madvise_large_batch.c

@@ -238,8 +238,7 @@ TEST_BEGIN(test_purge_more_than_one_batch_pages) {
 	}
 	for (int i = 0; i < HPA_PURGE_BATCH_MAX * 2 * (int)HUGEPAGE_PAGES;
 	    i++) {
-		hpa_dalloc(tsdn, shard, edatas[i],
+		hpa_dalloc(tsdn, shard, edatas[i], &deferred_work_generated);
-		    &deferred_work_generated);
 	}
 
 	hpa_shard_do_deferred_work(tsdn, shard);