Store a pointer to the free pages list that the binder allocator should use for a process inside of struct binder_alloc. This change allows binder allocator code to be tested and debugged deterministically while a system is using binder; i.e., without interfering with other binder processes and independently of the shrinker. This is necessary to convert the current binder_alloc_selftest into a kunit test that does not rely on hijacking an existing binder_proc to run.
A binder process's binder_alloc->freelist should not be changed after it is initialized. A sole exception is the process that runs the existing binder_alloc selftest. Its freelist can be temporarily replaced for the duration of the test because it runs as a single thread before any pages can be added to the global binder freelist, and the test frees every page it allocates before dropping the binder_selftest_lock. This exception allows the existing selftest to be used to check for regressions, but it will be dropped when the binder_alloc tests are converted to kunit in a subsequent patch in this series.
Acked-by: Carlos Llamas cmllamas@google.com Signed-off-by: Tiffany Yang ynaffit@google.com --- v4: * Collected tag --- drivers/android/binder_alloc.c | 25 +++++++---- drivers/android/binder_alloc.h | 3 +- drivers/android/binder_alloc_selftest.c | 59 ++++++++++++++++++++----- 3 files changed, 67 insertions(+), 20 deletions(-)
diff --git a/drivers/android/binder_alloc.c b/drivers/android/binder_alloc.c index fcfaf1b899c8..2e89f9127883 100644 --- a/drivers/android/binder_alloc.c +++ b/drivers/android/binder_alloc.c @@ -26,7 +26,7 @@ #include "binder_alloc.h" #include "binder_trace.h"
-struct list_lru binder_freelist; +static struct list_lru binder_freelist;
static DEFINE_MUTEX(binder_alloc_mmap_lock);
@@ -210,7 +210,7 @@ static void binder_lru_freelist_add(struct binder_alloc *alloc,
trace_binder_free_lru_start(alloc, index);
- ret = list_lru_add(&binder_freelist, + ret = list_lru_add(alloc->freelist, page_to_lru(page), page_to_nid(page), NULL); @@ -409,7 +409,7 @@ static void binder_lru_freelist_del(struct binder_alloc *alloc, if (page) { trace_binder_alloc_lru_start(alloc, index);
- on_lru = list_lru_del(&binder_freelist, + on_lru = list_lru_del(alloc->freelist, page_to_lru(page), page_to_nid(page), NULL); @@ -1007,7 +1007,7 @@ void binder_alloc_deferred_release(struct binder_alloc *alloc) if (!page) continue;
- on_lru = list_lru_del(&binder_freelist, + on_lru = list_lru_del(alloc->freelist, page_to_lru(page), page_to_nid(page), NULL); @@ -1229,6 +1229,17 @@ binder_shrink_scan(struct shrinker *shrink, struct shrink_control *sc)
static struct shrinker *binder_shrinker;
+static void __binder_alloc_init(struct binder_alloc *alloc, + struct list_lru *freelist) +{ + alloc->pid = current->group_leader->pid; + alloc->mm = current->mm; + mmgrab(alloc->mm); + mutex_init(&alloc->mutex); + INIT_LIST_HEAD(&alloc->buffers); + alloc->freelist = freelist; +} + /** * binder_alloc_init() - called by binder_open() for per-proc initialization * @alloc: binder_alloc for this proc @@ -1238,11 +1249,7 @@ static struct shrinker *binder_shrinker; */ void binder_alloc_init(struct binder_alloc *alloc) { - alloc->pid = current->group_leader->pid; - alloc->mm = current->mm; - mmgrab(alloc->mm); - mutex_init(&alloc->mutex); - INIT_LIST_HEAD(&alloc->buffers); + __binder_alloc_init(alloc, &binder_freelist); }
int binder_alloc_shrinker_init(void) diff --git a/drivers/android/binder_alloc.h b/drivers/android/binder_alloc.h index feecd7414241..aa05a9df1360 100644 --- a/drivers/android/binder_alloc.h +++ b/drivers/android/binder_alloc.h @@ -15,7 +15,6 @@ #include <linux/list_lru.h> #include <uapi/linux/android/binder.h>
-extern struct list_lru binder_freelist; struct binder_transaction;
/** @@ -91,6 +90,7 @@ static inline struct list_head *page_to_lru(struct page *p) * @free_async_space: VA space available for async buffers. This is * initialized at mmap time to 1/2 the full VA space * @pages: array of struct page * + * @freelist: lru list to use for free pages (invariant after init) * @buffer_size: size of address space specified via mmap * @pid: pid for associated binder_proc (invariant after init) * @pages_high: high watermark of offset in @pages @@ -113,6 +113,7 @@ struct binder_alloc { struct rb_root allocated_buffers; size_t free_async_space; struct page **pages; + struct list_lru *freelist; size_t buffer_size; int pid; size_t pages_high; diff --git a/drivers/android/binder_alloc_selftest.c b/drivers/android/binder_alloc_selftest.c index de5bd848d042..8b18b22aa3de 100644 --- a/drivers/android/binder_alloc_selftest.c +++ b/drivers/android/binder_alloc_selftest.c @@ -8,8 +8,9 @@
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
-#include <linux/mm_types.h> #include <linux/err.h> +#include <linux/list_lru.h> +#include <linux/mm_types.h> #include "binder_alloc.h"
#define BUFFER_NUM 5 @@ -18,6 +19,7 @@ static bool binder_selftest_run = true; static int binder_selftest_failures; static DEFINE_MUTEX(binder_selftest_lock); +static struct list_lru binder_selftest_freelist;
/** * enum buf_end_align_type - Page alignment of a buffer @@ -143,11 +145,6 @@ static void binder_selftest_free_buf(struct binder_alloc *alloc, binder_alloc_free_buf(alloc, buffers[seq[i]]);
for (i = 0; i <= (end - 1) / PAGE_SIZE; i++) { - /** - * Error message on a free page can be false positive - * if binder shrinker ran during binder_alloc_free_buf - * calls above. - */ if (list_empty(page_to_lru(alloc->pages[i]))) { pr_err_size_seq(sizes, seq); pr_err("expect lru but is %s at page index %d\n", @@ -162,8 +159,8 @@ static void binder_selftest_free_page(struct binder_alloc *alloc) int i; unsigned long count;
- while ((count = list_lru_count(&binder_freelist))) { - list_lru_walk(&binder_freelist, binder_alloc_free_page, + while ((count = list_lru_count(&binder_selftest_freelist))) { + list_lru_walk(&binder_selftest_freelist, binder_alloc_free_page, NULL, count); }
@@ -187,7 +184,7 @@ static void binder_selftest_alloc_free(struct binder_alloc *alloc,
/* Allocate from lru. */ binder_selftest_alloc_buf(alloc, buffers, sizes, seq); - if (list_lru_count(&binder_freelist)) + if (list_lru_count(&binder_selftest_freelist)) pr_err("lru list should be empty but is not\n");
binder_selftest_free_buf(alloc, buffers, sizes, seq, end); @@ -275,6 +272,20 @@ static void binder_selftest_alloc_offset(struct binder_alloc *alloc, } }
+int binder_selftest_alloc_get_page_count(struct binder_alloc *alloc) +{ + struct page *page; + int allocated = 0; + int i; + + for (i = 0; i < alloc->buffer_size / PAGE_SIZE; i++) { + page = alloc->pages[i]; + if (page) + allocated++; + } + return allocated; +} + /** * binder_selftest_alloc() - Test alloc and free of buffer pages. * @alloc: Pointer to alloc struct. @@ -286,6 +297,7 @@ static void binder_selftest_alloc_offset(struct binder_alloc *alloc, */ void binder_selftest_alloc(struct binder_alloc *alloc) { + struct list_lru *prev_freelist; size_t end_offset[BUFFER_NUM];
if (!binder_selftest_run) @@ -293,14 +305,41 @@ void binder_selftest_alloc(struct binder_alloc *alloc) mutex_lock(&binder_selftest_lock); if (!binder_selftest_run || !alloc->mapped) goto done; + + prev_freelist = alloc->freelist; + + /* + * It is not safe to modify this process's alloc->freelist if it has any + * pages on a freelist. Since the test runs before any binder ioctls can + * be dealt with, none of its pages should be allocated yet. + */ + if (binder_selftest_alloc_get_page_count(alloc)) { + pr_err("process has existing alloc state\n"); + goto cleanup; + } + + if (list_lru_init(&binder_selftest_freelist)) { + pr_err("failed to init test freelist\n"); + goto cleanup; + } + + alloc->freelist = &binder_selftest_freelist; + pr_info("STARTED\n"); binder_selftest_alloc_offset(alloc, end_offset, 0); - binder_selftest_run = false; if (binder_selftest_failures > 0) pr_info("%d tests FAILED\n", binder_selftest_failures); else pr_info("PASSED\n");
+ if (list_lru_count(&binder_selftest_freelist)) + pr_err("expect test freelist to be empty\n"); + +cleanup: + /* Even if we didn't run the test, it's no longer thread-safe. */ + binder_selftest_run = false; + alloc->freelist = prev_freelist; + list_lru_destroy(&binder_selftest_freelist); done: mutex_unlock(&binder_selftest_lock); }