Hi Al, Linus,
These patches make some changes to the kunit tests previously added for iov_iter testing, in particular adding support for testing UBUF/IOVEC iterators:
(1) Clean up a couple of checkpatch style complaints.
(2) Consolidate some repeated bits of code into helper functions and use the same struct to represent straight offset/address ranges and partial page lists.
(3) Add a function to set up a userspace VM, attach the VM to the kunit testing thread, create an anonymous file, stuff some pages into the file and map the file into the VM to act as a buffer that can be used with UBUF/IOVEC iterators.
I map an anonymous file with pages attached rather than using MAP_ANON so that I can check the pages obtained from iov_iter_extract_pages() without worrying about them changing due to swap, migrate, etc..
[?] Is this the best way to do things? Mirroring execve, it requires a number of extra core symbols to be exported. Should this be done in the core code?
(4) Add tests for copying into and out of UBUF and IOVEC iterators.
(5) Add tests for extracting pages from UBUF and IOVEC iterators.
(6) Add tests to benchmark copying 256MiB to UBUF, IOVEC, KVEC, BVEC and XARRAY iterators.
[!] Note that this requires 256MiB of memory for UBUF and IOVEC; the KVEC, BVEC and XARRAY benchmarking maps a single page multiple times. I might be able to shrink that if I can add the same page multiple times to the anon file's pagecache. I'm sure this is not recommended, but I might be able to get away with it for this particular application.
(7) Add a test to benchmark copying 256MiB through dynamically allocated 256-page bvecs to simulate bio construction.
Example benchmarks output:
iov_kunit_benchmark_ubuf: avg 26899 uS, stddev 142 uS iov_kunit_benchmark_iovec: avg 26897 uS, stddev 74 uS iov_kunit_benchmark_kvec: avg 2688 uS, stddev 35 uS iov_kunit_benchmark_bvec: avg 3139 uS, stddev 21 uS iov_kunit_benchmark_bvec_split: avg 3379 uS, stddev 15 uS iov_kunit_benchmark_xarray: avg 3582 uS, stddev 13 uS
I've pushed the patches here also:
https://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs.git/log/?h...
David
David Howells (9): iov_iter: Fix some checkpatch complaints in kunit tests iov_iter: Consolidate some of the repeated code into helpers iov_iter: Consolidate the test vector struct in the kunit tests iov_iter: Consolidate bvec pattern checking iov_iter: Create a function to prepare userspace VM for UBUF/IOVEC tests iov_iter: Add copy kunit tests for ITER_UBUF and ITER_IOVEC iov_iter: Add extract kunit tests for ITER_UBUF and ITER_IOVEC iov_iter: Add benchmarking kunit tests iov_iter: Add benchmarking kunit tests for UBUF/IOVEC
fs/anon_inodes.c | 1 + kernel/fork.c | 2 + lib/kunit_iov_iter.c | 1211 +++++++++++++++++++++++++++++++++++------- mm/mmap.c | 1 + mm/util.c | 1 + 5 files changed, 1024 insertions(+), 192 deletions(-)
Fix some checkpatch complaints in the new iov_iter kunit tests:
(1) Some lines had eight spaces instead of a tab at the start.
(2) Checkpatch doesn't like (void*)(unsigned long)0xnnnnnULL, so switch to using POISON_POINTER_DELTA plus an offset instead.
Reported-by: Johannes Thumshirn Johannes.Thumshirn@wdc.com Signed-off-by: David Howells dhowells@redhat.com cc: Christoph Hellwig hch@lst.de cc: Christian Brauner brauner@kernel.org cc: Jens Axboe axboe@kernel.dk cc: Al Viro viro@zeniv.linux.org.uk cc: David Hildenbrand david@redhat.com cc: John Hubbard jhubbard@nvidia.com cc: Brendan Higgins brendanhiggins@google.com cc: David Gow davidgow@google.com cc: linux-kselftest@vger.kernel.org cc: kunit-dev@googlegroups.com cc: linux-mm@kvack.org cc: linux-fsdevel@vger.kernel.org --- lib/kunit_iov_iter.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/lib/kunit_iov_iter.c b/lib/kunit_iov_iter.c index 859b67c4d697..4a6c0efd33f5 100644 --- a/lib/kunit_iov_iter.c +++ b/lib/kunit_iov_iter.c @@ -53,7 +53,7 @@ static void *__init iov_kunit_create_buffer(struct kunit *test, void *buffer;
pages = kunit_kcalloc(test, npages, sizeof(struct page *), GFP_KERNEL); - KUNIT_ASSERT_NOT_ERR_OR_NULL(test, pages); + KUNIT_ASSERT_NOT_ERR_OR_NULL(test, pages); *ppages = pages;
got = alloc_pages_bulk_array(GFP_KERNEL, npages, pages); @@ -63,7 +63,7 @@ static void *__init iov_kunit_create_buffer(struct kunit *test, }
buffer = vmap(pages, npages, VM_MAP | VM_MAP_PUT_PAGES, PAGE_KERNEL); - KUNIT_ASSERT_NOT_ERR_OR_NULL(test, buffer); + KUNIT_ASSERT_NOT_ERR_OR_NULL(test, buffer);
kunit_add_action_or_reset(test, iov_kunit_unmap, buffer); return buffer; @@ -548,7 +548,7 @@ static void __init iov_kunit_extract_pages_kvec(struct kunit *test) size_t offset0 = LONG_MAX;
for (i = 0; i < ARRAY_SIZE(pagelist); i++) - pagelist[i] = (void *)(unsigned long)0xaa55aa55aa55aa55ULL; + pagelist[i] = (void *)POISON_POINTER_DELTA + 0x5a;
len = iov_iter_extract_pages(&iter, &pages, 100 * 1024, ARRAY_SIZE(pagelist), 0, &offset0); @@ -626,7 +626,7 @@ static void __init iov_kunit_extract_pages_bvec(struct kunit *test) size_t offset0 = LONG_MAX;
for (i = 0; i < ARRAY_SIZE(pagelist); i++) - pagelist[i] = (void *)(unsigned long)0xaa55aa55aa55aa55ULL; + pagelist[i] = (void *)POISON_POINTER_DELTA + 0x5a;
len = iov_iter_extract_pages(&iter, &pages, 100 * 1024, ARRAY_SIZE(pagelist), 0, &offset0); @@ -709,7 +709,7 @@ static void __init iov_kunit_extract_pages_xarray(struct kunit *test) size_t offset0 = LONG_MAX;
for (i = 0; i < ARRAY_SIZE(pagelist); i++) - pagelist[i] = (void *)(unsigned long)0xaa55aa55aa55aa55ULL; + pagelist[i] = (void *)POISON_POINTER_DELTA + 0x5a;
len = iov_iter_extract_pages(&iter, &pages, 100 * 1024, ARRAY_SIZE(pagelist), 0, &offset0);
Consolidate some of the repeated code snippets into helper functions to reduce the line count.
Signed-off-by: David Howells dhowells@redhat.com cc: Christoph Hellwig hch@lst.de cc: Christian Brauner brauner@kernel.org cc: Jens Axboe axboe@kernel.dk cc: Al Viro viro@zeniv.linux.org.uk cc: David Hildenbrand david@redhat.com cc: John Hubbard jhubbard@nvidia.com cc: Brendan Higgins brendanhiggins@google.com cc: David Gow davidgow@google.com cc: linux-kselftest@vger.kernel.org cc: kunit-dev@googlegroups.com cc: linux-mm@kvack.org cc: linux-fsdevel@vger.kernel.org --- lib/kunit_iov_iter.c | 189 +++++++++++++++++++------------------------ 1 file changed, 84 insertions(+), 105 deletions(-)
diff --git a/lib/kunit_iov_iter.c b/lib/kunit_iov_iter.c index 4a6c0efd33f5..ee586eb652b4 100644 --- a/lib/kunit_iov_iter.c +++ b/lib/kunit_iov_iter.c @@ -19,18 +19,18 @@ MODULE_AUTHOR("David Howells dhowells@redhat.com"); MODULE_LICENSE("GPL");
struct kvec_test_range { - int from, to; + int page, from, to; };
static const struct kvec_test_range kvec_test_ranges[] = { - { 0x00002, 0x00002 }, - { 0x00027, 0x03000 }, - { 0x05193, 0x18794 }, - { 0x20000, 0x20000 }, - { 0x20000, 0x24000 }, - { 0x24000, 0x27001 }, - { 0x29000, 0xffffb }, - { 0xffffd, 0xffffe }, + { 0, 0x00002, 0x00002 }, + { 0, 0x00027, 0x03000 }, + { 0, 0x05193, 0x18794 }, + { 0, 0x20000, 0x20000 }, + { 0, 0x20000, 0x24000 }, + { 0, 0x24000, 0x27001 }, + { 0, 0x29000, 0xffffb }, + { 0, 0xffffd, 0xffffe }, { -1 } };
@@ -69,6 +69,57 @@ static void *__init iov_kunit_create_buffer(struct kunit *test, return buffer; }
+/* + * Build the reference pattern in the scratch buffer that we expect to see in + * the iterator buffer (ie. the result of copy *to*). + */ +static void iov_kunit_build_to_reference_pattern(struct kunit *test, u8 *scratch, + size_t bufsize, + const struct kvec_test_range *pr) +{ + int i, patt = 0; + + memset(scratch, 0, bufsize); + for (; pr->page >= 0; pr++) + for (i = pr->from; i < pr->to; i++) + scratch[i] = pattern(patt++); +} + +/* + * Build the reference pattern in the iterator buffer that we expect to see in + * the scratch buffer (ie. the result of copy *from*). + */ +static void iov_kunit_build_from_reference_pattern(struct kunit *test, u8 *buffer, + size_t bufsize, + const struct kvec_test_range *pr) +{ + size_t i = 0, j; + + memset(buffer, 0, bufsize); + for (; pr->page >= 0; pr++) { + for (j = pr->from; j < pr->to; j++) { + buffer[i++] = pattern(j); + if (i >= bufsize) + return; + } + } +} + +/* + * Compare two kernel buffers to see that they're the same. + */ +static void iov_kunit_check_pattern(struct kunit *test, const u8 *buffer, + const u8 *scratch, size_t bufsize) +{ + size_t i; + + for (i = 0; i < bufsize; i++) { + KUNIT_EXPECT_EQ_MSG(test, buffer[i], scratch[i], "at i=%x", i); + if (buffer[i] != scratch[i]) + return; + } +} + static void __init iov_kunit_load_kvec(struct kunit *test, struct iov_iter *iter, int dir, struct kvec *kvec, unsigned int kvmax, @@ -79,7 +130,7 @@ static void __init iov_kunit_load_kvec(struct kunit *test, int i;
for (i = 0; i < kvmax; i++, pr++) { - if (pr->from < 0) + if (pr->page < 0) break; KUNIT_ASSERT_GE(test, pr->to, pr->from); KUNIT_ASSERT_LE(test, pr->to, bufsize); @@ -97,13 +148,12 @@ static void __init iov_kunit_load_kvec(struct kunit *test, */ static void __init iov_kunit_copy_to_kvec(struct kunit *test) { - const struct kvec_test_range *pr; struct iov_iter iter; struct page **spages, **bpages; struct kvec kvec[8]; u8 *scratch, *buffer; size_t bufsize, npages, size, copied; - int i, patt; + int i;
bufsize = 0x100000; npages = bufsize / PAGE_SIZE; @@ -125,20 +175,8 @@ static void __init iov_kunit_copy_to_kvec(struct kunit *test) KUNIT_EXPECT_EQ(test, iter.count, 0); KUNIT_EXPECT_EQ(test, iter.nr_segs, 0);
- /* Build the expected image in the scratch buffer. */ - patt = 0; - memset(scratch, 0, bufsize); - for (pr = kvec_test_ranges; pr->from >= 0; pr++) - for (i = pr->from; i < pr->to; i++) - scratch[i] = pattern(patt++); - - /* Compare the images */ - for (i = 0; i < bufsize; i++) { - KUNIT_EXPECT_EQ_MSG(test, buffer[i], scratch[i], "at i=%x", i); - if (buffer[i] != scratch[i]) - return; - } - + iov_kunit_build_to_reference_pattern(test, scratch, bufsize, kvec_test_ranges); + iov_kunit_check_pattern(test, buffer, scratch, bufsize); KUNIT_SUCCEED(); }
@@ -147,13 +185,12 @@ static void __init iov_kunit_copy_to_kvec(struct kunit *test) */ static void __init iov_kunit_copy_from_kvec(struct kunit *test) { - const struct kvec_test_range *pr; struct iov_iter iter; struct page **spages, **bpages; struct kvec kvec[8]; u8 *scratch, *buffer; size_t bufsize, npages, size, copied; - int i, j; + int i;
bufsize = 0x100000; npages = bufsize / PAGE_SIZE; @@ -175,25 +212,8 @@ static void __init iov_kunit_copy_from_kvec(struct kunit *test) KUNIT_EXPECT_EQ(test, iter.count, 0); KUNIT_EXPECT_EQ(test, iter.nr_segs, 0);
- /* Build the expected image in the main buffer. */ - i = 0; - memset(buffer, 0, bufsize); - for (pr = kvec_test_ranges; pr->from >= 0; pr++) { - for (j = pr->from; j < pr->to; j++) { - buffer[i++] = pattern(j); - if (i >= bufsize) - goto stop; - } - } -stop: - - /* Compare the images */ - for (i = 0; i < bufsize; i++) { - KUNIT_EXPECT_EQ_MSG(test, scratch[i], buffer[i], "at i=%x", i); - if (scratch[i] != buffer[i]) - return; - } - + iov_kunit_build_from_reference_pattern(test, buffer, bufsize, kvec_test_ranges); + iov_kunit_check_pattern(test, buffer, scratch, bufsize); KUNIT_SUCCEED(); }
@@ -210,7 +230,7 @@ static const struct bvec_test_range bvec_test_ranges[] = { { 5, 0x0000, 0x1000 }, { 6, 0x0000, 0x0ffb }, { 6, 0x0ffd, 0x0ffe }, - { -1, -1, -1 } + { -1 } };
static void __init iov_kunit_load_bvec(struct kunit *test, @@ -225,7 +245,7 @@ static void __init iov_kunit_load_bvec(struct kunit *test, int i;
for (i = 0; i < bvmax; i++, pr++) { - if (pr->from < 0) + if (pr->page < 0) break; KUNIT_ASSERT_LT(test, pr->page, npages); KUNIT_ASSERT_LT(test, pr->page * PAGE_SIZE, bufsize); @@ -288,20 +308,14 @@ static void __init iov_kunit_copy_to_bvec(struct kunit *test) b = 0; patt = 0; memset(scratch, 0, bufsize); - for (pr = bvec_test_ranges; pr->from >= 0; pr++, b++) { + for (pr = bvec_test_ranges; pr->page >= 0; pr++, b++) { u8 *p = scratch + pr->page * PAGE_SIZE;
for (i = pr->from; i < pr->to; i++) p[i] = pattern(patt++); }
- /* Compare the images */ - for (i = 0; i < bufsize; i++) { - KUNIT_EXPECT_EQ_MSG(test, buffer[i], scratch[i], "at i=%x", i); - if (buffer[i] != scratch[i]) - return; - } - + iov_kunit_check_pattern(test, buffer, scratch, bufsize); KUNIT_SUCCEED(); }
@@ -341,7 +355,7 @@ static void __init iov_kunit_copy_from_bvec(struct kunit *test) /* Build the expected image in the main buffer. */ i = 0; memset(buffer, 0, bufsize); - for (pr = bvec_test_ranges; pr->from >= 0; pr++) { + for (pr = bvec_test_ranges; pr->page >= 0; pr++) { size_t patt = pr->page * PAGE_SIZE;
for (j = pr->from; j < pr->to; j++) { @@ -352,13 +366,7 @@ static void __init iov_kunit_copy_from_bvec(struct kunit *test) } stop:
- /* Compare the images */ - for (i = 0; i < bufsize; i++) { - KUNIT_EXPECT_EQ_MSG(test, scratch[i], buffer[i], "at i=%x", i); - if (scratch[i] != buffer[i]) - return; - } - + iov_kunit_check_pattern(test, buffer, scratch, bufsize); KUNIT_SUCCEED(); }
@@ -409,7 +417,7 @@ static void __init iov_kunit_copy_to_xarray(struct kunit *test) struct page **spages, **bpages; u8 *scratch, *buffer; size_t bufsize, npages, size, copied; - int i, patt; + int i;
bufsize = 0x100000; npages = bufsize / PAGE_SIZE; @@ -426,7 +434,7 @@ static void __init iov_kunit_copy_to_xarray(struct kunit *test) iov_kunit_load_xarray(test, &iter, READ, xarray, bpages, npages);
i = 0; - for (pr = kvec_test_ranges; pr->from >= 0; pr++) { + for (pr = kvec_test_ranges; pr->page >= 0; pr++) { size = pr->to - pr->from; KUNIT_ASSERT_LE(test, pr->to, bufsize);
@@ -439,20 +447,8 @@ static void __init iov_kunit_copy_to_xarray(struct kunit *test) i += size; }
- /* Build the expected image in the scratch buffer. */ - patt = 0; - memset(scratch, 0, bufsize); - for (pr = kvec_test_ranges; pr->from >= 0; pr++) - for (i = pr->from; i < pr->to; i++) - scratch[i] = pattern(patt++); - - /* Compare the images */ - for (i = 0; i < bufsize; i++) { - KUNIT_EXPECT_EQ_MSG(test, buffer[i], scratch[i], "at i=%x", i); - if (buffer[i] != scratch[i]) - return; - } - + iov_kunit_build_to_reference_pattern(test, scratch, bufsize, kvec_test_ranges); + iov_kunit_check_pattern(test, buffer, scratch, bufsize); KUNIT_SUCCEED(); }
@@ -467,7 +463,7 @@ static void __init iov_kunit_copy_from_xarray(struct kunit *test) struct page **spages, **bpages; u8 *scratch, *buffer; size_t bufsize, npages, size, copied; - int i, j; + int i;
bufsize = 0x100000; npages = bufsize / PAGE_SIZE; @@ -484,7 +480,7 @@ static void __init iov_kunit_copy_from_xarray(struct kunit *test) iov_kunit_load_xarray(test, &iter, READ, xarray, bpages, npages);
i = 0; - for (pr = kvec_test_ranges; pr->from >= 0; pr++) { + for (pr = kvec_test_ranges; pr->page >= 0; pr++) { size = pr->to - pr->from; KUNIT_ASSERT_LE(test, pr->to, bufsize);
@@ -497,25 +493,8 @@ static void __init iov_kunit_copy_from_xarray(struct kunit *test) i += size; }
- /* Build the expected image in the main buffer. */ - i = 0; - memset(buffer, 0, bufsize); - for (pr = kvec_test_ranges; pr->from >= 0; pr++) { - for (j = pr->from; j < pr->to; j++) { - buffer[i++] = pattern(j); - if (i >= bufsize) - goto stop; - } - } -stop: - - /* Compare the images */ - for (i = 0; i < bufsize; i++) { - KUNIT_EXPECT_EQ_MSG(test, scratch[i], buffer[i], "at i=%x", i); - if (scratch[i] != buffer[i]) - return; - } - + iov_kunit_build_from_reference_pattern(test, buffer, bufsize, kvec_test_ranges); + iov_kunit_check_pattern(test, buffer, scratch, bufsize); KUNIT_SUCCEED(); }
@@ -573,7 +552,7 @@ static void __init iov_kunit_extract_pages_kvec(struct kunit *test) while (from == pr->to) { pr++; from = pr->from; - if (from < 0) + if (pr->page < 0) goto stop; } ix = from / PAGE_SIZE; @@ -651,7 +630,7 @@ static void __init iov_kunit_extract_pages_bvec(struct kunit *test) while (from == pr->to) { pr++; from = pr->from; - if (from < 0) + if (pr->page < 0) goto stop; } ix = pr->page + from / PAGE_SIZE; @@ -698,7 +677,7 @@ static void __init iov_kunit_extract_pages_xarray(struct kunit *test) iov_kunit_create_buffer(test, &bpages, npages); iov_kunit_load_xarray(test, &iter, READ, xarray, bpages, npages);
- for (pr = kvec_test_ranges; pr->from >= 0; pr++) { + for (pr = kvec_test_ranges; pr->page >= 0; pr++) { from = pr->from; size = pr->to - from; KUNIT_ASSERT_LE(test, pr->to, bufsize);
Consolidate the test vector struct in the kunit tests so that the bvec pattern check helpers can share with the kvec check helpers.
Signed-off-by: David Howells dhowells@redhat.com cc: Christoph Hellwig hch@lst.de cc: Christian Brauner brauner@kernel.org cc: Jens Axboe axboe@kernel.dk cc: Al Viro viro@zeniv.linux.org.uk cc: David Hildenbrand david@redhat.com cc: John Hubbard jhubbard@nvidia.com cc: Brendan Higgins brendanhiggins@google.com cc: David Gow davidgow@google.com cc: linux-kselftest@vger.kernel.org cc: kunit-dev@googlegroups.com cc: linux-mm@kvack.org cc: linux-fsdevel@vger.kernel.org --- lib/kunit_iov_iter.c | 90 ++++++++++++++++++++++++-------------------- 1 file changed, 50 insertions(+), 40 deletions(-)
diff --git a/lib/kunit_iov_iter.c b/lib/kunit_iov_iter.c index ee586eb652b4..4925ca37cde6 100644 --- a/lib/kunit_iov_iter.c +++ b/lib/kunit_iov_iter.c @@ -18,22 +18,46 @@ MODULE_DESCRIPTION("iov_iter testing"); MODULE_AUTHOR("David Howells dhowells@redhat.com"); MODULE_LICENSE("GPL");
-struct kvec_test_range { +struct iov_kunit_range { int page, from, to; };
-static const struct kvec_test_range kvec_test_ranges[] = { - { 0, 0x00002, 0x00002 }, - { 0, 0x00027, 0x03000 }, - { 0, 0x05193, 0x18794 }, - { 0, 0x20000, 0x20000 }, - { 0, 0x20000, 0x24000 }, - { 0, 0x24000, 0x27001 }, - { 0, 0x29000, 0xffffb }, - { 0, 0xffffd, 0xffffe }, +/* + * Ranges that to use in tests where we have address/offset ranges to play + * with (ie. KVEC) or where we have a single blob that we can copy + * arbitrary chunks of (ie. XARRAY). + */ +static const struct iov_kunit_range kvec_test_ranges[] = { + { 0, 0x00002, 0x00002 }, /* Start with an empty range */ + { 0, 0x00027, 0x03000 }, /* Midpage to page end */ + { 0, 0x05193, 0x18794 }, /* Midpage to midpage */ + { 0, 0x20000, 0x20000 }, /* Empty range in the middle */ + { 0, 0x20000, 0x24000 }, /* Page start to page end */ + { 0, 0x24000, 0x27001 }, /* Page end to midpage */ + { 0, 0x29000, 0xffffb }, /* Page start to midpage */ + { 0, 0xffffd, 0xffffe }, /* Almost contig to last, ending in same page */ { -1 } };
+/* + * Ranges that to use in tests where we have a list of partial pages to + * play with (ie. BVEC). + */ +static const struct iov_kunit_range bvec_test_ranges[] = { + { 0, 0x0002, 0x0002 }, /* Start with an empty range */ + { 1, 0x0027, 0x0893 }, /* Random part of page */ + { 2, 0x0193, 0x0794 }, /* Random part of page */ + { 3, 0x0000, 0x1000 }, /* Full page */ + { 4, 0x0000, 0x1000 }, /* Full page logically contig to last */ + { 5, 0x0000, 0x1000 }, /* Full page logically contig to last */ + { 6, 0x0000, 0x0ffb }, /* Part page logically contig to last */ + { 6, 0x0ffd, 0x0ffe }, /* Part of prev page, but not quite contig */ + { -1 } +}; + +/* + * The pattern to fill with. + */ static inline u8 pattern(unsigned long x) { return x & 0xff; @@ -44,6 +68,9 @@ static void iov_kunit_unmap(void *data) vunmap(data); }
+/* + * Create a buffer out of some pages and return a vmap'd pointer to it. + */ static void *__init iov_kunit_create_buffer(struct kunit *test, struct page ***ppages, size_t npages) @@ -75,7 +102,7 @@ static void *__init iov_kunit_create_buffer(struct kunit *test, */ static void iov_kunit_build_to_reference_pattern(struct kunit *test, u8 *scratch, size_t bufsize, - const struct kvec_test_range *pr) + const struct iov_kunit_range *pr) { int i, patt = 0;
@@ -91,7 +118,7 @@ static void iov_kunit_build_to_reference_pattern(struct kunit *test, u8 *scratch */ static void iov_kunit_build_from_reference_pattern(struct kunit *test, u8 *buffer, size_t bufsize, - const struct kvec_test_range *pr) + const struct iov_kunit_range *pr) { size_t i = 0, j;
@@ -124,7 +151,7 @@ static void __init iov_kunit_load_kvec(struct kunit *test, struct iov_iter *iter, int dir, struct kvec *kvec, unsigned int kvmax, void *buffer, size_t bufsize, - const struct kvec_test_range *pr) + const struct iov_kunit_range *pr) { size_t size = 0; int i; @@ -217,28 +244,12 @@ static void __init iov_kunit_copy_from_kvec(struct kunit *test) KUNIT_SUCCEED(); }
-struct bvec_test_range { - int page, from, to; -}; - -static const struct bvec_test_range bvec_test_ranges[] = { - { 0, 0x0002, 0x0002 }, - { 1, 0x0027, 0x0893 }, - { 2, 0x0193, 0x0794 }, - { 3, 0x0000, 0x1000 }, - { 4, 0x0000, 0x1000 }, - { 5, 0x0000, 0x1000 }, - { 6, 0x0000, 0x0ffb }, - { 6, 0x0ffd, 0x0ffe }, - { -1 } -}; - static void __init iov_kunit_load_bvec(struct kunit *test, struct iov_iter *iter, int dir, struct bio_vec *bvec, unsigned int bvmax, struct page **pages, size_t npages, size_t bufsize, - const struct bvec_test_range *pr) + const struct iov_kunit_range *pr) { struct page *can_merge = NULL, *page; size_t size = 0; @@ -276,13 +287,13 @@ static void __init iov_kunit_load_bvec(struct kunit *test, */ static void __init iov_kunit_copy_to_bvec(struct kunit *test) { - const struct bvec_test_range *pr; + const struct iov_kunit_range *pr; struct iov_iter iter; struct bio_vec bvec[8]; struct page **spages, **bpages; u8 *scratch, *buffer; size_t bufsize, npages, size, copied; - int i, b, patt; + int i, patt;
bufsize = 0x100000; npages = bufsize / PAGE_SIZE; @@ -305,10 +316,9 @@ static void __init iov_kunit_copy_to_bvec(struct kunit *test) KUNIT_EXPECT_EQ(test, iter.nr_segs, 0);
/* Build the expected image in the scratch buffer. */ - b = 0; patt = 0; memset(scratch, 0, bufsize); - for (pr = bvec_test_ranges; pr->page >= 0; pr++, b++) { + for (pr = bvec_test_ranges; pr->page >= 0; pr++) { u8 *p = scratch + pr->page * PAGE_SIZE;
for (i = pr->from; i < pr->to; i++) @@ -324,7 +334,7 @@ static void __init iov_kunit_copy_to_bvec(struct kunit *test) */ static void __init iov_kunit_copy_from_bvec(struct kunit *test) { - const struct bvec_test_range *pr; + const struct iov_kunit_range *pr; struct iov_iter iter; struct bio_vec bvec[8]; struct page **spages, **bpages; @@ -411,7 +421,7 @@ static struct xarray *iov_kunit_create_xarray(struct kunit *test) */ static void __init iov_kunit_copy_to_xarray(struct kunit *test) { - const struct kvec_test_range *pr; + const struct iov_kunit_range *pr; struct iov_iter iter; struct xarray *xarray; struct page **spages, **bpages; @@ -457,7 +467,7 @@ static void __init iov_kunit_copy_to_xarray(struct kunit *test) */ static void __init iov_kunit_copy_from_xarray(struct kunit *test) { - const struct kvec_test_range *pr; + const struct iov_kunit_range *pr; struct iov_iter iter; struct xarray *xarray; struct page **spages, **bpages; @@ -503,7 +513,7 @@ static void __init iov_kunit_copy_from_xarray(struct kunit *test) */ static void __init iov_kunit_extract_pages_kvec(struct kunit *test) { - const struct kvec_test_range *pr; + const struct iov_kunit_range *pr; struct iov_iter iter; struct page **bpages, *pagelist[8], **pages = pagelist; struct kvec kvec[8]; @@ -583,7 +593,7 @@ static void __init iov_kunit_extract_pages_kvec(struct kunit *test) */ static void __init iov_kunit_extract_pages_bvec(struct kunit *test) { - const struct bvec_test_range *pr; + const struct iov_kunit_range *pr; struct iov_iter iter; struct page **bpages, *pagelist[8], **pages = pagelist; struct bio_vec bvec[8]; @@ -661,7 +671,7 @@ static void __init iov_kunit_extract_pages_bvec(struct kunit *test) */ static void __init iov_kunit_extract_pages_xarray(struct kunit *test) { - const struct kvec_test_range *pr; + const struct iov_kunit_range *pr; struct iov_iter iter; struct xarray *xarray; struct page **bpages, *pagelist[8], **pages = pagelist;
Make the BVEC-testing functions use the consolidated pattern checking functions to reduce the amount of duplicated code.
Signed-off-by: David Howells dhowells@redhat.com cc: Christoph Hellwig hch@lst.de cc: Christian Brauner brauner@kernel.org cc: Jens Axboe axboe@kernel.dk cc: Al Viro viro@zeniv.linux.org.uk cc: David Hildenbrand david@redhat.com cc: John Hubbard jhubbard@nvidia.com cc: Brendan Higgins brendanhiggins@google.com cc: David Gow davidgow@google.com cc: linux-kselftest@vger.kernel.org cc: kunit-dev@googlegroups.com cc: linux-mm@kvack.org cc: linux-fsdevel@vger.kernel.org --- lib/kunit_iov_iter.c | 42 +++++++++++------------------------------- 1 file changed, 11 insertions(+), 31 deletions(-)
diff --git a/lib/kunit_iov_iter.c b/lib/kunit_iov_iter.c index 4925ca37cde6..eb86371b67d0 100644 --- a/lib/kunit_iov_iter.c +++ b/lib/kunit_iov_iter.c @@ -107,9 +107,11 @@ static void iov_kunit_build_to_reference_pattern(struct kunit *test, u8 *scratch int i, patt = 0;
memset(scratch, 0, bufsize); - for (; pr->page >= 0; pr++) + for (; pr->page >= 0; pr++) { + u8 *p = scratch + pr->page * PAGE_SIZE; for (i = pr->from; i < pr->to; i++) - scratch[i] = pattern(patt++); + p[i] = pattern(patt++); + } }
/* @@ -124,8 +126,10 @@ static void iov_kunit_build_from_reference_pattern(struct kunit *test, u8 *buffe
memset(buffer, 0, bufsize); for (; pr->page >= 0; pr++) { + size_t patt = pr->page * PAGE_SIZE; + for (j = pr->from; j < pr->to; j++) { - buffer[i++] = pattern(j); + buffer[i++] = pattern(patt + j); if (i >= bufsize) return; } @@ -287,13 +291,12 @@ static void __init iov_kunit_load_bvec(struct kunit *test, */ static void __init iov_kunit_copy_to_bvec(struct kunit *test) { - const struct iov_kunit_range *pr; struct iov_iter iter; struct bio_vec bvec[8]; struct page **spages, **bpages; u8 *scratch, *buffer; size_t bufsize, npages, size, copied; - int i, patt; + int i;
bufsize = 0x100000; npages = bufsize / PAGE_SIZE; @@ -315,16 +318,7 @@ static void __init iov_kunit_copy_to_bvec(struct kunit *test) KUNIT_EXPECT_EQ(test, iter.count, 0); KUNIT_EXPECT_EQ(test, iter.nr_segs, 0);
- /* Build the expected image in the scratch buffer. */ - patt = 0; - memset(scratch, 0, bufsize); - for (pr = bvec_test_ranges; pr->page >= 0; pr++) { - u8 *p = scratch + pr->page * PAGE_SIZE; - - for (i = pr->from; i < pr->to; i++) - p[i] = pattern(patt++); - } - + iov_kunit_build_to_reference_pattern(test, scratch, bufsize, bvec_test_ranges); iov_kunit_check_pattern(test, buffer, scratch, bufsize); KUNIT_SUCCEED(); } @@ -334,13 +328,12 @@ static void __init iov_kunit_copy_to_bvec(struct kunit *test) */ static void __init iov_kunit_copy_from_bvec(struct kunit *test) { - const struct iov_kunit_range *pr; struct iov_iter iter; struct bio_vec bvec[8]; struct page **spages, **bpages; u8 *scratch, *buffer; size_t bufsize, npages, size, copied; - int i, j; + int i;
bufsize = 0x100000; npages = bufsize / PAGE_SIZE; @@ -362,20 +355,7 @@ static void __init iov_kunit_copy_from_bvec(struct kunit *test) KUNIT_EXPECT_EQ(test, iter.count, 0); KUNIT_EXPECT_EQ(test, iter.nr_segs, 0);
- /* Build the expected image in the main buffer. */ - i = 0; - memset(buffer, 0, bufsize); - for (pr = bvec_test_ranges; pr->page >= 0; pr++) { - size_t patt = pr->page * PAGE_SIZE; - - for (j = pr->from; j < pr->to; j++) { - buffer[i++] = pattern(patt + j); - if (i >= bufsize) - goto stop; - } - } -stop: - + iov_kunit_build_from_reference_pattern(test, buffer, bufsize, bvec_test_ranges); iov_kunit_check_pattern(test, buffer, scratch, bufsize); KUNIT_SUCCEED(); }
Create a function to set up a userspace VM for the kunit testing thread and set up a buffer within it such that ITER_UBUF and ITER_IOVEC tests can be performed.
Note that this requires current->mm to point to a sufficiently set up mm_struct. This is done by partially mirroring what execve does.
The following steps are performed:
(1) Allocate an mm_struct and pick an arch layout (required to set mm->get_unmapped_area).
(2) Create an empty "stack" VMA so that the VMA maple tree is set up and won't cause a crash in the maple tree code later. We don't actually care about the stack as we're not going to actually execute userspace.
(3) Create an anon file and attach a bunch of folios to it so that the requested number of pages are accessible.
(4) Make the kthread use the mm. This must be done before mmap is called.
(5) Shared-mmap the anon file into the allocated mm_struct.
This requires access to otherwise unexported core symbols: mm_alloc(), vm_area_alloc(), insert_vm_struct() arch_pick_mmap_layout() and anon_inode_getfile_secure(), which I've exported _GPL.
[?] Would it be better if this were done in core and not in a module?
Signed-off-by: David Howells dhowells@redhat.com cc: Andrew Morton akpm@linux-foundation.org cc: Christoph Hellwig hch@lst.de cc: Christian Brauner brauner@kernel.org cc: Jens Axboe axboe@kernel.dk cc: Al Viro viro@zeniv.linux.org.uk cc: Matthew Wilcox willy@infradead.org cc: David Hildenbrand david@redhat.com cc: John Hubbard jhubbard@nvidia.com cc: Brendan Higgins brendanhiggins@google.com cc: David Gow davidgow@google.com cc: linux-mm@kvack.org cc: linux-fsdevel@vger.kernel.org cc: linux-kselftest@vger.kernel.org cc: kunit-dev@googlegroups.com --- fs/anon_inodes.c | 1 + kernel/fork.c | 2 + lib/kunit_iov_iter.c | 158 +++++++++++++++++++++++++++++++++++++++++++ mm/mmap.c | 1 + mm/util.c | 1 + 5 files changed, 163 insertions(+)
diff --git a/fs/anon_inodes.c b/fs/anon_inodes.c index 24192a7667ed..4190336180ee 100644 --- a/fs/anon_inodes.c +++ b/fs/anon_inodes.c @@ -176,6 +176,7 @@ struct file *anon_inode_getfile_secure(const char *name, return __anon_inode_getfile(name, fops, priv, flags, context_inode, true); } +EXPORT_SYMBOL_GPL(anon_inode_getfile_secure);
static int __anon_inode_getfd(const char *name, const struct file_operations *fops, diff --git a/kernel/fork.c b/kernel/fork.c index 3b6d20dfb9a8..9ab604574400 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -494,6 +494,7 @@ struct vm_area_struct *vm_area_alloc(struct mm_struct *mm)
return vma; } +EXPORT_SYMBOL_GPL(vm_area_alloc);
struct vm_area_struct *vm_area_dup(struct vm_area_struct *orig) { @@ -1337,6 +1338,7 @@ struct mm_struct *mm_alloc(void) memset(mm, 0, sizeof(*mm)); return mm_init(mm, current, current_user_ns()); } +EXPORT_SYMBOL_GPL(mm_alloc);
static inline void __mmput(struct mm_struct *mm) { diff --git a/lib/kunit_iov_iter.c b/lib/kunit_iov_iter.c index eb86371b67d0..3353bca9c40f 100644 --- a/lib/kunit_iov_iter.c +++ b/lib/kunit_iov_iter.c @@ -10,6 +10,12 @@ #include <linux/module.h> #include <linux/vmalloc.h> #include <linux/mm.h> +#include <linux/pagemap.h> +#include <linux/mman.h> +#include <linux/file.h> +#include <linux/kthread.h> +#include <linux/anon_inodes.h> +#include <linux/writeback.h> #include <linux/uio.h> #include <linux/bvec.h> #include <kunit/test.h> @@ -68,6 +74,20 @@ static void iov_kunit_unmap(void *data) vunmap(data); }
+static void iov_kunit_mmdrop(void *data) +{ + struct mm_struct *mm = data; + + if (current->mm == mm) + kthread_unuse_mm(mm); + mmdrop(mm); +} + +static void iov_kunit_fput(void *data) +{ + fput(data); +} + /* * Create a buffer out of some pages and return a vmap'd pointer to it. */ @@ -96,6 +116,20 @@ static void *__init iov_kunit_create_buffer(struct kunit *test, return buffer; }
+static void iov_kunit_fill_user_buf(struct kunit *test, + u8 __user *buffer, size_t bufsize) +{ + size_t i; + int err; + + for (i = 0; i < bufsize; i++) { + err = put_user(pattern(i), &buffer[i]); + KUNIT_EXPECT_EQ(test, err, 0); + if (test->status == KUNIT_FAILURE) + return; + } +} + /* * Build the reference pattern in the scratch buffer that we expect to see in * the iterator buffer (ie. the result of copy *to*). @@ -151,6 +185,130 @@ static void iov_kunit_check_pattern(struct kunit *test, const u8 *buffer, } }
+/* + * Compare a user and a scratch buffer to see that they're the same. + */ +static void iov_kunit_check_user_pattern(struct kunit *test, const u8 __user *buffer, + const u8 *scratch, size_t bufsize) +{ + size_t i; + int err; + u8 c; + + for (i = 0; i < bufsize; i++) { + err = get_user(c, &buffer[i]); + KUNIT_EXPECT_EQ(test, err, 0); + KUNIT_EXPECT_EQ_MSG(test, c, scratch[i], "at i=%x", i); + if (c != scratch[i]) + return; + } +} + +static const struct file_operations iov_kunit_user_file_fops = { + .mmap = generic_file_mmap, +}; + +static int iov_kunit_user_file_read_folio(struct file *file, struct folio *folio) +{ + folio_mark_uptodate(folio); + folio_unlock(folio); + return 0; +} + +static const struct address_space_operations iov_kunit_user_file_aops = { + .read_folio = iov_kunit_user_file_read_folio, + .dirty_folio = filemap_dirty_folio, +}; + +/* + * Create an anonymous file and attach a bunch of pages to it. We can then use + * this in mmap() and check the pages against it when doing extraction tests. + */ +static struct file *iov_kunit_create_file(struct kunit *test, size_t npages, + struct page ***ppages) +{ + struct folio *folio; + struct file *file; + struct page **pages = NULL; + size_t i; + + if (ppages) { + pages = kunit_kcalloc(test, npages, sizeof(struct page *), GFP_KERNEL); + KUNIT_ASSERT_NOT_ERR_OR_NULL(test, pages); + *ppages = pages; + } + + file = anon_inode_getfile_secure("kunit-iov-test", + &iov_kunit_user_file_fops, + NULL, O_RDWR, NULL); + KUNIT_ASSERT_NOT_ERR_OR_NULL(test, file); + kunit_add_action_or_reset(test, iov_kunit_fput, file); + file->f_mapping->a_ops = &iov_kunit_user_file_aops; + + i_size_write(file_inode(file), npages * PAGE_SIZE); + for (i = 0; i < npages; i++) { + folio = filemap_grab_folio(file->f_mapping, i); + KUNIT_ASSERT_NOT_ERR_OR_NULL(test, folio); + if (pages) + *pages++ = folio_page(folio, 0); + folio_unlock(folio); + folio_put(folio); + } + + return file; +} + +static u8 __user *__init iov_kunit_create_user_buf(struct kunit *test, + size_t npages, + struct page ***ppages) +{ + struct rlimit rlim_stack = { + .rlim_cur = LONG_MAX, + .rlim_max = LONG_MAX, + }; + struct vm_area_struct *vma; + struct mm_struct *mm; + struct file *file; + u8 __user *buffer; + int ret; + + KUNIT_ASSERT_NULL(test, current->mm); + + mm = mm_alloc(); + KUNIT_ASSERT_NOT_NULL(test, mm); + kunit_add_action_or_reset(test, iov_kunit_mmdrop, mm); + arch_pick_mmap_layout(mm, &rlim_stack); + + vma = vm_area_alloc(mm); + KUNIT_ASSERT_NOT_NULL(test, vma); + vma_set_anonymous(vma); + + /* + * Place the stack at the largest stack address the architecture + * supports. Later, we'll move this to an appropriate place. We don't + * use STACK_TOP because that can depend on attributes which aren't + * configured yet. + */ + vma->vm_end = STACK_TOP_MAX; + vma->vm_start = vma->vm_end - PAGE_SIZE; + vm_flags_init(vma, VM_SOFTDIRTY | VM_STACK_FLAGS | VM_STACK_INCOMPLETE_SETUP); + vma->vm_page_prot = vm_get_page_prot(vma->vm_flags); + + ret = insert_vm_struct(mm, vma); + KUNIT_ASSERT_EQ(test, ret, 0); + + mm->stack_vm = mm->total_vm = 1; + + file = iov_kunit_create_file(test, npages, ppages); + + kthread_use_mm(mm); + buffer = (u8 __user *)vm_mmap(file, 0, PAGE_SIZE * npages, + PROT_READ | PROT_WRITE, + MAP_SHARED, 0); + KUNIT_ASSERT_NOT_ERR_OR_NULL(test, buffer); + return buffer; +} + static void __init iov_kunit_load_kvec(struct kunit *test, struct iov_iter *iter, int dir, struct kvec *kvec, unsigned int kvmax, diff --git a/mm/mmap.c b/mm/mmap.c index b56a7f0c9f85..2ea4a98a2cab 100644 --- a/mm/mmap.c +++ b/mm/mmap.c @@ -3284,6 +3284,7 @@ int insert_vm_struct(struct mm_struct *mm, struct vm_area_struct *vma)
return 0; } +EXPORT_SYMBOL_GPL(insert_vm_struct);
/* * Copy the vma structure to a new location in the same mm, diff --git a/mm/util.c b/mm/util.c index 8cbbfd3a3d59..a393a308607c 100644 --- a/mm/util.c +++ b/mm/util.c @@ -455,6 +455,7 @@ void arch_pick_mmap_layout(struct mm_struct *mm, struct rlimit *rlim_stack) mm->get_unmapped_area = arch_get_unmapped_area; } #endif +EXPORT_SYMBOL_GPL(arch_pick_mmap_layout);
/** * __account_locked_vm - account locked pages to an mm's locked_vm
Add copy kunit tests for ITER_UBUF- and ITER_IOVEC-type iterators. This attaches a userspace VM with a mapped file in it temporarily to the test thread.
Signed-off-by: David Howells dhowells@redhat.com cc: Andrew Morton akpm@linux-foundation.org cc: Christoph Hellwig hch@lst.de cc: Christian Brauner brauner@kernel.org cc: Jens Axboe axboe@kernel.dk cc: Al Viro viro@zeniv.linux.org.uk cc: Matthew Wilcox willy@infradead.org cc: David Hildenbrand david@redhat.com cc: John Hubbard jhubbard@nvidia.com cc: Brendan Higgins brendanhiggins@google.com cc: David Gow davidgow@google.com cc: linux-mm@kvack.org cc: linux-fsdevel@vger.kernel.org cc: linux-kselftest@vger.kernel.org cc: kunit-dev@googlegroups.com --- lib/kunit_iov_iter.c | 200 +++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 200 insertions(+)
diff --git a/lib/kunit_iov_iter.c b/lib/kunit_iov_iter.c index 3353bca9c40f..78f566ebd4a6 100644 --- a/lib/kunit_iov_iter.c +++ b/lib/kunit_iov_iter.c @@ -309,6 +309,202 @@ static u8 __user *__init iov_kunit_create_user_buf(struct kunit *test, return buffer; }
+/* + * Test copying to an ITER_UBUF-type iterator. + */ +static void __init iov_kunit_copy_to_ubuf(struct kunit *test) +{ + const struct iov_kunit_range *pr; + struct iov_iter iter; + struct page **spages; + u8 __user *buffer; + u8 *scratch; + ssize_t uncleared; + size_t bufsize, npages, size, copied; + int i; + + bufsize = 0x100000; + npages = bufsize / PAGE_SIZE; + + scratch = iov_kunit_create_buffer(test, &spages, npages); + for (i = 0; i < bufsize; i++) + scratch[i] = pattern(i); + + buffer = iov_kunit_create_user_buf(test, npages, NULL); + uncleared = clear_user(buffer, bufsize); + KUNIT_EXPECT_EQ(test, uncleared, 0); + if (uncleared) + return; + + i = 0; + for (pr = kvec_test_ranges; pr->page >= 0; pr++) { + size = pr->to - pr->from; + KUNIT_ASSERT_LE(test, pr->to, bufsize); + + iov_iter_ubuf(&iter, ITER_DEST, buffer + pr->from, size); + copied = copy_to_iter(scratch + i, size, &iter); + + KUNIT_EXPECT_EQ(test, copied, size); + KUNIT_EXPECT_EQ(test, iter.count, 0); + KUNIT_EXPECT_EQ(test, iter.iov_offset, size); + if (test->status == KUNIT_FAILURE) + break; + i += size; + } + + iov_kunit_build_to_reference_pattern(test, scratch, bufsize, kvec_test_ranges); + iov_kunit_check_user_pattern(test, buffer, scratch, bufsize); + KUNIT_SUCCEED(); +} + +/* + * Test copying from an ITER_UBUF-type iterator. + */ +static void __init iov_kunit_copy_from_ubuf(struct kunit *test) +{ + const struct iov_kunit_range *pr; + struct iov_iter iter; + struct page **spages; + u8 __user *buffer; + u8 *scratch, *reference; + size_t bufsize, npages, size, copied; + int i; + + bufsize = 0x100000; + npages = bufsize / PAGE_SIZE; + + buffer = iov_kunit_create_user_buf(test, npages, NULL); + iov_kunit_fill_user_buf(test, buffer, bufsize); + + scratch = iov_kunit_create_buffer(test, &spages, npages); + memset(scratch, 0, bufsize); + + reference = iov_kunit_create_buffer(test, &spages, npages); + + i = 0; + for (pr = kvec_test_ranges; pr->page >= 0; pr++) { + size = pr->to - pr->from; + KUNIT_ASSERT_LE(test, pr->to, bufsize); + + iov_iter_ubuf(&iter, ITER_SOURCE, buffer + pr->from, size); + copied = copy_from_iter(scratch + i, size, &iter); + + KUNIT_EXPECT_EQ(test, copied, size); + KUNIT_EXPECT_EQ(test, iter.count, 0); + KUNIT_EXPECT_EQ(test, iter.iov_offset, size); + if (test->status == KUNIT_FAILURE) + break; + i += size; + } + + iov_kunit_build_from_reference_pattern(test, reference, bufsize, kvec_test_ranges); + iov_kunit_check_pattern(test, scratch, reference, bufsize); + KUNIT_SUCCEED(); +} + +static void __init iov_kunit_load_iovec(struct kunit *test, + struct iov_iter *iter, int dir, + struct iovec *iov, unsigned int iovmax, + u8 __user *buffer, size_t bufsize, + const struct iov_kunit_range *pr) +{ + size_t size = 0; + int i; + + for (i = 0; i < iovmax; i++, pr++) { + if (pr->page < 0) + break; + KUNIT_ASSERT_GE(test, pr->to, pr->from); + KUNIT_ASSERT_LE(test, pr->to, bufsize); + iov[i].iov_base = buffer + pr->from; + iov[i].iov_len = pr->to - pr->from; + size += pr->to - pr->from; + } + KUNIT_ASSERT_LE(test, size, bufsize); + + iov_iter_init(iter, dir, iov, i, size); +} + +/* + * Test copying to an ITER_IOVEC-type iterator. + */ +static void __init iov_kunit_copy_to_iovec(struct kunit *test) +{ + struct iov_iter iter; + struct page **spages; + struct iovec iov[8]; + u8 __user *buffer; + u8 *scratch; + ssize_t uncleared; + size_t bufsize, npages, size, copied; + int i; + + bufsize = 0x100000; + npages = bufsize / PAGE_SIZE; + + scratch = iov_kunit_create_buffer(test, &spages, npages); + for (i = 0; i < bufsize; i++) + scratch[i] = pattern(i); + + buffer = iov_kunit_create_user_buf(test, npages, NULL); + uncleared = clear_user(buffer, bufsize); + KUNIT_EXPECT_EQ(test, uncleared, 0); + if (uncleared) + return; + + iov_kunit_load_iovec(test, &iter, ITER_DEST, iov, ARRAY_SIZE(iov), + buffer, bufsize, kvec_test_ranges); + size = iter.count; + + copied = copy_to_iter(scratch, size, &iter); + + KUNIT_EXPECT_EQ(test, copied, size); + KUNIT_EXPECT_EQ(test, iter.count, 0); + KUNIT_EXPECT_EQ(test, iter.nr_segs, 0); + + iov_kunit_build_to_reference_pattern(test, scratch, bufsize, kvec_test_ranges); + iov_kunit_check_user_pattern(test, buffer, scratch, bufsize); + KUNIT_SUCCEED(); +} + +/* + * Test copying from an ITER_IOVEC-type iterator. + */ +static void __init iov_kunit_copy_from_iovec(struct kunit *test) +{ + struct iov_iter iter; + struct page **spages; + struct iovec iov[8]; + u8 __user *buffer; + u8 *scratch, *reference; + size_t bufsize, npages, size, copied; + + bufsize = 0x100000; + npages = bufsize / PAGE_SIZE; + + buffer = iov_kunit_create_user_buf(test, npages, NULL); + iov_kunit_fill_user_buf(test, buffer, bufsize); + + scratch = iov_kunit_create_buffer(test, &spages, npages); + memset(scratch, 0, bufsize); + + reference = iov_kunit_create_buffer(test, &spages, npages); + + iov_kunit_load_iovec(test, &iter, ITER_SOURCE, iov, ARRAY_SIZE(iov), + buffer, bufsize, kvec_test_ranges); + size = iter.count; + + copied = copy_from_iter(scratch, size, &iter); + + KUNIT_EXPECT_EQ(test, copied, size); + KUNIT_EXPECT_EQ(test, iter.count, 0); + KUNIT_EXPECT_EQ(test, iter.nr_segs, 0); + + iov_kunit_build_from_reference_pattern(test, reference, bufsize, kvec_test_ranges); + iov_kunit_check_pattern(test, reference, scratch, bufsize); + KUNIT_SUCCEED(); +} + static void __init iov_kunit_load_kvec(struct kunit *test, struct iov_iter *iter, int dir, struct kvec *kvec, unsigned int kvmax, @@ -884,6 +1080,10 @@ static void __init iov_kunit_extract_pages_xarray(struct kunit *test) }
static struct kunit_case __refdata iov_kunit_cases[] = { + KUNIT_CASE(iov_kunit_copy_to_ubuf), + KUNIT_CASE(iov_kunit_copy_from_ubuf), + KUNIT_CASE(iov_kunit_copy_to_iovec), + KUNIT_CASE(iov_kunit_copy_from_iovec), KUNIT_CASE(iov_kunit_copy_to_kvec), KUNIT_CASE(iov_kunit_copy_from_kvec), KUNIT_CASE(iov_kunit_copy_to_bvec),
Add extraction kunit tests for ITER_UBUF- and ITER_IOVEC-type iterators. This attaches a userspace VM with a mapped file in it temporarily to the test thread.
[!] Note that this requires the kernel thread running the test to obtain and deploy an mm_struct so that a user-side buffer can be created with mmap - basically it has to emulated part of execve(). Doing so requires access to additional core symbols: mm_alloc(), vm_area_alloc(), insert_vm_struct() and arch_pick_mmap_layout(). See the iov_kunit_create_user_buf() function added in the patch.
Signed-off-by: David Howells dhowells@redhat.com cc: Andrew Morton akpm@linux-foundation.org cc: Christoph Hellwig hch@lst.de cc: Christian Brauner brauner@kernel.org cc: Jens Axboe axboe@kernel.dk cc: Al Viro viro@zeniv.linux.org.uk cc: Matthew Wilcox willy@infradead.org cc: David Hildenbrand david@redhat.com cc: John Hubbard jhubbard@nvidia.com cc: Brendan Higgins brendanhiggins@google.com cc: David Gow davidgow@google.com cc: linux-mm@kvack.org cc: linux-fsdevel@vger.kernel.org cc: linux-kselftest@vger.kernel.org cc: kunit-dev@googlegroups.com --- lib/kunit_iov_iter.c | 164 +++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 164 insertions(+)
diff --git a/lib/kunit_iov_iter.c b/lib/kunit_iov_iter.c index 78f566ebd4a6..0b7961bbab62 100644 --- a/lib/kunit_iov_iter.c +++ b/lib/kunit_iov_iter.c @@ -842,6 +842,168 @@ static void __init iov_kunit_copy_from_xarray(struct kunit *test) KUNIT_SUCCEED(); }
+/* + * Test the extraction of ITER_UBUF-type iterators. + */ +static void __init iov_kunit_extract_pages_ubuf(struct kunit *test) +{ + const struct iov_kunit_range *pr; + struct iov_iter iter; + struct page **bpages, *pagelist[8], **pages = pagelist; + ssize_t len; + size_t bufsize, size = 0, npages; + int i, from; + u8 __user *buffer; + + bufsize = 0x100000; + npages = bufsize / PAGE_SIZE; + + buffer = iov_kunit_create_user_buf(test, npages, &bpages); + + for (pr = kvec_test_ranges; pr->page >= 0; pr++) { + from = pr->from; + size = pr->to - from; + KUNIT_ASSERT_LE(test, pr->to, bufsize); + + iov_iter_ubuf(&iter, ITER_SOURCE, buffer + pr->from, size); + + do { + size_t offset0 = LONG_MAX; + + for (i = 0; i < ARRAY_SIZE(pagelist); i++) + pagelist[i] = (void *)POISON_POINTER_DELTA + 0x5a; + + len = iov_iter_extract_pages(&iter, &pages, 100 * 1024, + ARRAY_SIZE(pagelist), 0, &offset0); + KUNIT_EXPECT_GE(test, len, 0); + if (len < 0) + break; + KUNIT_EXPECT_LE(test, len, size); + KUNIT_EXPECT_EQ(test, iter.count, size - len); + if (len == 0) + break; + size -= len; + KUNIT_EXPECT_GE(test, (ssize_t)offset0, 0); + KUNIT_EXPECT_LT(test, offset0, PAGE_SIZE); + + /* We're only checking the page pointers */ + unpin_user_pages(pages, (offset0 + len) / PAGE_SIZE); + + for (i = 0; i < ARRAY_SIZE(pagelist); i++) { + struct page *p; + ssize_t part = min_t(ssize_t, len, PAGE_SIZE - offset0); + int ix; + + KUNIT_ASSERT_GE(test, part, 0); + ix = from / PAGE_SIZE; + KUNIT_ASSERT_LT(test, ix, npages); + p = bpages[ix]; + KUNIT_EXPECT_PTR_EQ(test, pagelist[i], p); + KUNIT_EXPECT_EQ(test, offset0, from % PAGE_SIZE); + from += part; + len -= part; + KUNIT_ASSERT_GE(test, len, 0); + if (len == 0) + break; + offset0 = 0; + } + + if (test->status == KUNIT_FAILURE) + goto stop; + } while (iov_iter_count(&iter) > 0); + + KUNIT_EXPECT_EQ(test, size, 0); + KUNIT_EXPECT_EQ(test, iter.count, 0); + KUNIT_EXPECT_EQ(test, iter.iov_offset, pr->to - pr->from); + } + +stop: + KUNIT_SUCCEED(); +} + +/* + * Test the extraction of ITER_IOVEC-type iterators. + */ +static void __init iov_kunit_extract_pages_iovec(struct kunit *test) +{ + const struct iov_kunit_range *pr; + struct iov_iter iter; + struct iovec iov[8]; + struct page **bpages, *pagelist[8], **pages = pagelist; + ssize_t len; + size_t bufsize, size = 0, npages; + int i, from; + u8 __user *buffer; + + bufsize = 0x100000; + npages = bufsize / PAGE_SIZE; + + buffer = iov_kunit_create_user_buf(test, npages, &bpages); + + iov_kunit_load_iovec(test, &iter, ITER_SOURCE, iov, ARRAY_SIZE(iov), + buffer, bufsize, kvec_test_ranges); + size = iter.count; + + pr = kvec_test_ranges; + from = pr->from; + do { + size_t offset0 = LONG_MAX; + + for (i = 0; i < ARRAY_SIZE(pagelist); i++) + pagelist[i] = (void *)POISON_POINTER_DELTA + 0x5a; + + len = iov_iter_extract_pages(&iter, &pages, 100 * 1024, + ARRAY_SIZE(pagelist), 0, &offset0); + KUNIT_EXPECT_GE(test, len, 0); + if (len < 0) + break; + KUNIT_EXPECT_LE(test, len, size); + KUNIT_EXPECT_EQ(test, iter.count, size - len); + if (len == 0) + break; + size -= len; + KUNIT_EXPECT_GE(test, (ssize_t)offset0, 0); + KUNIT_EXPECT_LT(test, offset0, PAGE_SIZE); + + /* We're only checking the page pointers */ + unpin_user_pages(pages, (offset0 + len) / PAGE_SIZE); + + for (i = 0; i < ARRAY_SIZE(pagelist); i++) { + struct page *p; + ssize_t part = min_t(ssize_t, len, PAGE_SIZE - offset0); + int ix; + + KUNIT_ASSERT_GE(test, part, 0); + while (from == pr->to) { + pr++; + from = pr->from; + if (pr->page < 0) + goto stop; + } + + ix = from / PAGE_SIZE; + KUNIT_ASSERT_LT(test, ix, npages); + p = bpages[ix]; + KUNIT_EXPECT_PTR_EQ(test, pagelist[i], p); + KUNIT_EXPECT_EQ(test, offset0, from % PAGE_SIZE); + from += part; + len -= part; + KUNIT_ASSERT_GE(test, len, 0); + if (len == 0) + break; + offset0 = 0; + } + + if (test->status == KUNIT_FAILURE) + break; + } while (iov_iter_count(&iter) > 0); + +stop: + KUNIT_EXPECT_EQ(test, size, 0); + KUNIT_EXPECT_EQ(test, iter.count, 0); + KUNIT_SUCCEED(); +} + /* * Test the extraction of ITER_KVEC-type iterators. */ @@ -1090,6 +1252,8 @@ static struct kunit_case __refdata iov_kunit_cases[] = { KUNIT_CASE(iov_kunit_copy_from_bvec), KUNIT_CASE(iov_kunit_copy_to_xarray), KUNIT_CASE(iov_kunit_copy_from_xarray), + KUNIT_CASE(iov_kunit_extract_pages_ubuf), + KUNIT_CASE(iov_kunit_extract_pages_iovec), KUNIT_CASE(iov_kunit_extract_pages_kvec), KUNIT_CASE(iov_kunit_extract_pages_bvec), KUNIT_CASE(iov_kunit_extract_pages_xarray),
Add kunit tests to benchmark 256MiB copies to a KVEC iterator, a BVEC iterator, an XARRAY iterator and to a loop that allocates 256-page BVECs and fills them in (similar to a maximal bio struct being set up).
Signed-off-by: David Howells dhowells@redhat.com cc: Christoph Hellwig hch@lst.de cc: Christian Brauner brauner@kernel.org cc: Jens Axboe axboe@kernel.dk cc: Al Viro viro@zeniv.linux.org.uk cc: David Hildenbrand david@redhat.com cc: John Hubbard jhubbard@nvidia.com cc: Brendan Higgins brendanhiggins@google.com cc: David Gow davidgow@google.com cc: linux-kselftest@vger.kernel.org cc: kunit-dev@googlegroups.com cc: linux-fsdevel@vger.kernel.org cc: linux-mm@kvack.org --- lib/kunit_iov_iter.c | 251 +++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 251 insertions(+)
diff --git a/lib/kunit_iov_iter.c b/lib/kunit_iov_iter.c index 0b7961bbab62..f8d0cd6a2923 100644 --- a/lib/kunit_iov_iter.c +++ b/lib/kunit_iov_iter.c @@ -1241,6 +1241,253 @@ static void __init iov_kunit_extract_pages_xarray(struct kunit *test) KUNIT_SUCCEED(); }
+static void iov_kunit_free_page(void *data) +{ + __free_page(data); +} + +#define IOV_KUNIT_NR_SAMPLES 16 +static void __init iov_kunit_benchmark_print_stats(struct kunit *test, + unsigned int *samples) +{ + unsigned long long sumsq = 0; + unsigned long total = 0, mean, stddev; + unsigned int n = IOV_KUNIT_NR_SAMPLES; + int i; + + //for (i = 0; i < n; i++) + // kunit_info(test, "run %x: %u uS\n", i, samples[i]); + + /* Ignore the 0th sample as that may include extra overhead such as + * setting up PTEs. + */ + samples++; + n--; + for (i = 0; i < n; i++) + total += samples[i]; + mean = total / n; + + for (i = 0; i < n; i++) { + long s = samples[i] - mean; + + sumsq += s * s; + } + stddev = int_sqrt64(sumsq); + + kunit_info(test, "avg %lu uS, stddev %lu uS\n", mean, stddev); +} + +/* + * Create a source buffer for benchmarking. + */ +static void *__init iov_kunit_create_source(struct kunit *test, size_t npages) +{ + struct page *page, **pages; + void *scratch; + size_t i; + + /* Allocate a page and tile it repeatedly in the buffer. */ + page = alloc_page(GFP_KERNEL); + KUNIT_ASSERT_NOT_NULL(test, page); + kunit_add_action_or_reset(test, iov_kunit_free_page, page); + + pages = kunit_kmalloc_array(test, npages, sizeof(pages[0]), GFP_KERNEL); + KUNIT_ASSERT_NOT_NULL(test, pages); + for (i = 0; i < npages; i++) { + pages[i] = page; + get_page(page); + } + + scratch = vmap(pages, npages, VM_MAP | VM_MAP_PUT_PAGES, PAGE_KERNEL); + KUNIT_ASSERT_NOT_NULL(test, scratch); + kunit_add_action_or_reset(test, iov_kunit_unmap, scratch); + return scratch; +} + +/* + * Time copying 256MiB through an ITER_KVEC. + */ +static void __init iov_kunit_benchmark_kvec(struct kunit *test) +{ + struct iov_iter iter; + struct kvec kvec[8]; + unsigned int samples[IOV_KUNIT_NR_SAMPLES]; + ktime_t a, b; + ssize_t copied; + size_t size = 256 * 1024 * 1024, npages = size / PAGE_SIZE, part; + void *scratch, *buffer; + int i; + + /* Allocate a huge buffer and populate it with pages. */ + buffer = iov_kunit_create_source(test, npages); + + /* Create a single large buffer to copy to/from. */ + scratch = iov_kunit_create_source(test, npages); + + /* Split the target over a number of kvecs */ + copied = 0; + for (i = 0; i < ARRAY_SIZE(kvec); i++) { + part = size / ARRAY_SIZE(kvec); + kvec[i].iov_base = buffer + copied; + kvec[i].iov_len = part; + copied += part; + } + kvec[i - 1].iov_len += size - part; + + /* Perform and time a bunch of copies. */ + kunit_info(test, "Benchmarking copy_to_iter() over KVEC:\n"); + for (i = 0; i < IOV_KUNIT_NR_SAMPLES; i++) { + iov_iter_kvec(&iter, ITER_DEST, kvec, ARRAY_SIZE(kvec), size); + + a = ktime_get_real(); + copied = copy_to_iter(scratch, size, &iter); + b = ktime_get_real(); + KUNIT_EXPECT_EQ(test, copied, size); + samples[i] = ktime_to_us(ktime_sub(b, a)); + } + + iov_kunit_benchmark_print_stats(test, samples); + KUNIT_SUCCEED(); +} + +/* + * Time copying 256MiB through an ITER_BVEC. + */ +static void __init iov_kunit_benchmark_bvec(struct kunit *test) +{ + struct iov_iter iter; + struct bio_vec *bvec; + struct page *page; + unsigned int samples[IOV_KUNIT_NR_SAMPLES]; + ktime_t a, b; + ssize_t copied; + size_t size = 256 * 1024 * 1024, npages = size / PAGE_SIZE; + void *scratch; + int i; + + /* Allocate a page and tile it repeatedly in the buffer. */ + page = alloc_page(GFP_KERNEL); + KUNIT_ASSERT_NOT_NULL(test, page); + kunit_add_action_or_reset(test, iov_kunit_free_page, page); + + bvec = kunit_kmalloc_array(test, npages, sizeof(bvec[0]), GFP_KERNEL); + KUNIT_ASSERT_NOT_NULL(test, bvec); + for (i = 0; i < npages; i++) + bvec_set_page(&bvec[i], page, PAGE_SIZE, 0); + + /* Create a single large buffer to copy to/from. */ + scratch = iov_kunit_create_source(test, npages); + + /* Perform and time a bunch of copies. */ + kunit_info(test, "Benchmarking copy_to_iter() over BVEC:\n"); + for (i = 0; i < IOV_KUNIT_NR_SAMPLES; i++) { + iov_iter_bvec(&iter, ITER_DEST, bvec, npages, size); + a = ktime_get_real(); + copied = copy_to_iter(scratch, size, &iter); + b = ktime_get_real(); + KUNIT_EXPECT_EQ(test, copied, size); + samples[i] = ktime_to_us(ktime_sub(b, a)); + } + + iov_kunit_benchmark_print_stats(test, samples); + KUNIT_SUCCEED(); +} + +/* + * Time copying 256MiB through an ITER_BVEC in 256 page chunks. + */ +static void __init iov_kunit_benchmark_bvec_split(struct kunit *test) +{ + struct iov_iter iter; + struct bio_vec *bvec; + struct page *page; + unsigned int samples[IOV_KUNIT_NR_SAMPLES]; + ktime_t a, b; + ssize_t copied; + size_t size, npages = 64; + void *scratch; + int i, j; + + /* Allocate a page and tile it repeatedly in the buffer. */ + page = alloc_page(GFP_KERNEL); + KUNIT_ASSERT_NOT_NULL(test, page); + kunit_add_action_or_reset(test, iov_kunit_free_page, page); + + /* Create a single large buffer to copy to/from. */ + scratch = iov_kunit_create_source(test, npages); + + /* Perform and time a bunch of copies. */ + kunit_info(test, "Benchmarking copy_to_iter() over BVEC:\n"); + for (i = 0; i < IOV_KUNIT_NR_SAMPLES; i++) { + size = 256 * 1024 * 1024; + a = ktime_get_real(); + do { + size_t part = min(size, npages * PAGE_SIZE); + + bvec = kunit_kmalloc_array(test, npages, sizeof(bvec[0]), GFP_KERNEL); + KUNIT_ASSERT_NOT_NULL(test, bvec); + for (j = 0; j < npages; j++) + bvec_set_page(&bvec[j], page, PAGE_SIZE, 0); + + iov_iter_bvec(&iter, ITER_DEST, bvec, npages, part); + copied = copy_to_iter(scratch, part, &iter); + KUNIT_EXPECT_EQ(test, copied, part); + size -= part; + } while (size > 0); + b = ktime_get_real(); + samples[i] = ktime_to_us(ktime_sub(b, a)); + } + + iov_kunit_benchmark_print_stats(test, samples); + KUNIT_SUCCEED(); +} + +/* + * Time copying 256MiB through an ITER_XARRAY. + */ +static void __init iov_kunit_benchmark_xarray(struct kunit *test) +{ + struct iov_iter iter; + struct xarray *xarray; + struct page *page; + unsigned int samples[IOV_KUNIT_NR_SAMPLES]; + ktime_t a, b; + ssize_t copied; + size_t size = 256 * 1024 * 1024, npages = size / PAGE_SIZE; + void *scratch; + int i; + + /* Allocate a page and tile it repeatedly in the buffer. */ + page = alloc_page(GFP_KERNEL); + KUNIT_ASSERT_NOT_NULL(test, page); + kunit_add_action_or_reset(test, iov_kunit_free_page, page); + + xarray = iov_kunit_create_xarray(test); + + for (i = 0; i < npages; i++) { + void *x = xa_store(xarray, i, page, GFP_KERNEL); + + KUNIT_ASSERT_FALSE(test, xa_is_err(x)); + } + + /* Create a single large buffer to copy to/from. */ + scratch = iov_kunit_create_source(test, npages); + + /* Perform and time a bunch of copies. */ + kunit_info(test, "Benchmarking copy_to_iter() over XARRAY:\n"); + for (i = 0; i < IOV_KUNIT_NR_SAMPLES; i++) { + iov_iter_xarray(&iter, ITER_DEST, xarray, 0, size); + a = ktime_get_real(); + copied = copy_to_iter(scratch, size, &iter); + b = ktime_get_real(); + KUNIT_EXPECT_EQ(test, copied, size); + samples[i] = ktime_to_us(ktime_sub(b, a)); + } + + iov_kunit_benchmark_print_stats(test, samples); + KUNIT_SUCCEED(); +} + static struct kunit_case __refdata iov_kunit_cases[] = { KUNIT_CASE(iov_kunit_copy_to_ubuf), KUNIT_CASE(iov_kunit_copy_from_ubuf), @@ -1257,6 +1504,10 @@ static struct kunit_case __refdata iov_kunit_cases[] = { KUNIT_CASE(iov_kunit_extract_pages_kvec), KUNIT_CASE(iov_kunit_extract_pages_bvec), KUNIT_CASE(iov_kunit_extract_pages_xarray), + KUNIT_CASE(iov_kunit_benchmark_kvec), + KUNIT_CASE(iov_kunit_benchmark_bvec), + KUNIT_CASE(iov_kunit_benchmark_bvec_split), + KUNIT_CASE(iov_kunit_benchmark_xarray), {} };
Add kunit tests to benchmark 256MiB copies to a UBUF iterator and an IOVEC iterator. This attaches a userspace VM with a mapped file in it temporarily to the test thread.
Signed-off-by: David Howells dhowells@redhat.com cc: Andrew Morton akpm@linux-foundation.org cc: Christoph Hellwig hch@lst.de cc: Christian Brauner brauner@kernel.org cc: Jens Axboe axboe@kernel.dk cc: Al Viro viro@zeniv.linux.org.uk cc: Matthew Wilcox willy@infradead.org cc: David Hildenbrand david@redhat.com cc: John Hubbard jhubbard@nvidia.com cc: Brendan Higgins brendanhiggins@google.com cc: David Gow davidgow@google.com cc: linux-kselftest@vger.kernel.org cc: kunit-dev@googlegroups.com cc: linux-mm@kvack.org cc: linux-fsdevel@vger.kernel.org --- lib/kunit_iov_iter.c | 85 ++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 85 insertions(+)
diff --git a/lib/kunit_iov_iter.c b/lib/kunit_iov_iter.c index f8d0cd6a2923..cc9c64663a73 100644 --- a/lib/kunit_iov_iter.c +++ b/lib/kunit_iov_iter.c @@ -1304,6 +1304,89 @@ static void *__init iov_kunit_create_source(struct kunit *test, size_t npages) return scratch; }
+/* + * Time copying 256MiB through an ITER_UBUF. + */ +static void __init iov_kunit_benchmark_ubuf(struct kunit *test) +{ + struct iov_iter iter; + unsigned int samples[IOV_KUNIT_NR_SAMPLES]; + ktime_t a, b; + ssize_t copied; + size_t size = 256 * 1024 * 1024, npages = size / PAGE_SIZE; + void *scratch; + int i; + u8 __user *buffer; + + /* Allocate a huge buffer and populate it with pages. */ + buffer = iov_kunit_create_user_buf(test, npages, NULL); + + /* Create a single large buffer to copy to/from. */ + scratch = iov_kunit_create_source(test, npages); + + /* Perform and time a bunch of copies. */ + kunit_info(test, "Benchmarking copy_to_iter() over UBUF:\n"); + for (i = 0; i < IOV_KUNIT_NR_SAMPLES; i++) { + iov_iter_ubuf(&iter, ITER_DEST, buffer, size); + + a = ktime_get_real(); + copied = copy_to_iter(scratch, size, &iter); + b = ktime_get_real(); + KUNIT_EXPECT_EQ(test, copied, size); + samples[i] = ktime_to_us(ktime_sub(b, a)); + } + + iov_kunit_benchmark_print_stats(test, samples); + KUNIT_SUCCEED(); +} + +/* + * Time copying 256MiB through an ITER_IOVEC. + */ +static void __init iov_kunit_benchmark_iovec(struct kunit *test) +{ + struct iov_iter iter; + struct iovec iov[8]; + unsigned int samples[IOV_KUNIT_NR_SAMPLES]; + ktime_t a, b; + ssize_t copied; + size_t size = 256 * 1024 * 1024, npages = size / PAGE_SIZE, part; + void *scratch; + int i; + u8 __user *buffer; + + /* Allocate a huge buffer and populate it with pages. */ + buffer = iov_kunit_create_user_buf(test, npages, NULL); + + /* Create a single large buffer to copy to/from. */ + scratch = iov_kunit_create_source(test, npages); + + /* Split the target over a number of iovecs */ + copied = 0; + for (i = 0; i < ARRAY_SIZE(iov); i++) { + part = size / ARRAY_SIZE(iov); + iov[i].iov_base = buffer + copied; + iov[i].iov_len = part; + copied += part; + } + iov[i - 1].iov_len += size - part; + + /* Perform and time a bunch of copies. */ + kunit_info(test, "Benchmarking copy_to_iter() over IOVEC:\n"); + for (i = 0; i < IOV_KUNIT_NR_SAMPLES; i++) { + iov_iter_init(&iter, ITER_DEST, iov, ARRAY_SIZE(iov), size); + + a = ktime_get_real(); + copied = copy_to_iter(scratch, size, &iter); + b = ktime_get_real(); + KUNIT_EXPECT_EQ(test, copied, size); + samples[i] = ktime_to_us(ktime_sub(b, a)); + } + + iov_kunit_benchmark_print_stats(test, samples); + KUNIT_SUCCEED(); +} + /* * Time copying 256MiB through an ITER_KVEC. */ @@ -1504,6 +1587,8 @@ static struct kunit_case __refdata iov_kunit_cases[] = { KUNIT_CASE(iov_kunit_extract_pages_kvec), KUNIT_CASE(iov_kunit_extract_pages_bvec), KUNIT_CASE(iov_kunit_extract_pages_xarray), + KUNIT_CASE(iov_kunit_benchmark_ubuf), + KUNIT_CASE(iov_kunit_benchmark_iovec), KUNIT_CASE(iov_kunit_benchmark_kvec), KUNIT_CASE(iov_kunit_benchmark_bvec), KUNIT_CASE(iov_kunit_benchmark_bvec_split),
From: David Howells
Sent: 14 September 2023 23:15
Add kunit tests to benchmark 256MiB copies to a UBUF iterator and an IOVEC iterator. This attaches a userspace VM with a mapped file in it temporarily to the test thread.
Isn't that going to be completely dominated by the cache fills from memory?
I'd have thought you'd need to use something with a lot of small fragments so that the iteration code dominates the copy.
Some measurements can be made using readv() and writev() on /dev/zero and /dev/null.
David
- Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
David Laight David.Laight@ACULAB.COM wrote:
Add kunit tests to benchmark 256MiB copies to a UBUF iterator and an IOVEC iterator. This attaches a userspace VM with a mapped file in it temporarily to the test thread.
Isn't that going to be completely dominated by the cache fills from memory?
Yes... but it should be consistent in the amount of time that consumes since no device drivers are involved. I can try adding the same folio to the anon_file multiple times - it might work especially if I don't put the pages on the LRU (if that's even possible) - but I wanted separate pages for the extraction test.
I'd have thought you'd need to use something with a lot of small fragments so that the iteration code dominates the copy.
That would actually be a separate benchmark case which I should try also.
Some measurements can be made using readv() and writev() on /dev/zero and /dev/null.
Forget /dev/null; that doesn't actually engage any iteration code. The same for writing to /dev/zero. Reading from /dev/zero does its own iteration thing rather than using iterate_and_advance(), presumably because it checks for signals and resched.
David
From: David Howells
Sent: 15 September 2023 11:10
David Laight David.Laight@ACULAB.COM wrote:
Add kunit tests to benchmark 256MiB copies to a UBUF iterator and an IOVEC iterator. This attaches a userspace VM with a mapped file in it temporarily to the test thread.
Isn't that going to be completely dominated by the cache fills from memory?
Yes... but it should be consistent in the amount of time that consumes since no device drivers are involved. I can try adding the same folio to the anon_file multiple times - it might work especially if I don't put the pages on the LRU (if that's even possible) - but I wanted separate pages for the extraction test.
You could also just not do the copy! Although you need (say) asm volatile("\n",:::"memory") to stop it all being completely optimised away. That might show up a difference in the 'out_of_line' test where 15% on top on the data copies is massive - it may be that the data cache behaviour is very different for the two cases.
...
Some measurements can be made using readv() and writev() on /dev/zero and /dev/null.
Forget /dev/null; that doesn't actually engage any iteration code. The same for writing to /dev/zero. Reading from /dev/zero does its own iteration thing rather than using iterate_and_advance(), presumably because it checks for signals and resched.
Using /dev/null does exercise the 'copy iov from user' code. Last time I looked at that the 32bit compat code was faster than the 64bit code on x86!
David
- Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
David Laight David.Laight@ACULAB.COM wrote:
Some measurements can be made using readv() and writev() on /dev/zero and /dev/null.
Forget /dev/null; that doesn't actually engage any iteration code. The same for writing to /dev/zero. Reading from /dev/zero does its own iteration thing rather than using iterate_and_advance(), presumably because it checks for signals and resched.
Using /dev/null does exercise the 'copy iov from user' code.
Ummm.... Not really:
static ssize_t read_null(struct file *file, char __user *buf, size_t count, loff_t *ppos) { return 0; }
static ssize_t write_null(struct file *file, const char __user *buf, size_t count, loff_t *ppos) { return count; }
static ssize_t read_iter_null(struct kiocb *iocb, struct iov_iter *to) { return 0; }
static ssize_t write_iter_null(struct kiocb *iocb, struct iov_iter *from) { size_t count = iov_iter_count(from); iov_iter_advance(from, count); return count; }
David
From: David Howells
Sent: 15 September 2023 12:23
David Laight David.Laight@ACULAB.COM wrote:
Some measurements can be made using readv() and writev() on /dev/zero and /dev/null.
Forget /dev/null; that doesn't actually engage any iteration code. The same for writing to /dev/zero. Reading from /dev/zero does its own iteration thing rather than using iterate_and_advance(), presumably because it checks for signals and resched.
Using /dev/null does exercise the 'copy iov from user' code.
Ummm.... Not really:
I was thinking of import_iovec() - or whatever its current name is.
That really needs a single structure that contains the iov_iter and the cache[] (which the caller pretty much always allocates in the same place). Fiddling with that is ok until you find what io_uring does. Then it all gets entirely horrid.
David
- Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
David Laight David.Laight@ACULAB.COM wrote:
I was thinking of import_iovec() - or whatever its current name is.
That doesn't actually access the buffer described by the iovec[].
That really needs a single structure that contains the iov_iter and the cache[] (which the caller pretty much always allocates in the same place).
cache[]?
Fiddling with that is ok until you find what io_uring does. Then it all gets entirely horrid.
That statement sounds like back-of-the-OLS-T-shirt material ;-)
David
From: David Howells
Sent: 15 September 2023 13:36
David Laight David.Laight@ACULAB.COM wrote:
I was thinking of import_iovec() - or whatever its current name is.
That doesn't actually access the buffer described by the iovec[].
That really needs a single structure that contains the iov_iter and the cache[] (which the caller pretty much always allocates in the same place).
cache[]?
Ah it is usually called iovstack[].
That is the code that reads the iovec[] from user. For small counts there is an on-stack cache[], for large counts it has call kmalloc(). So when the io completes you have to free the allocated buffer.
A canonical example is:
static ssize_t vfs_readv(struct file *file, const struct iovec __user *vec, unsigned long vlen, loff_t *pos, rwf_t flags) { struct iovec iovstack[UIO_FASTIOV]; struct iovec *iov = iovstack; struct iov_iter iter; ssize_t ret;
ret = import_iovec(ITER_DEST, vec, vlen, ARRAY_SIZE(iovstack), &iov, &iter); if (ret >= 0) { ret = do_iter_read(file, &iter, pos, flags); kfree(iov); }
return ret; }
If 'iter' and 'iovstack' are put together in a structure the calling sequence becomes much less annoying. The kfree() can (probably) check iter.iovec != iovsatack (as an inline).
But io_uring manages to allocate the iov_iter and iovstack[] in entirely different places - and then copies them about.
David
- Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
David Laight David.Laight@ACULAB.COM wrote:
You could also just not do the copy! Although you need (say) asm volatile("\n",:::"memory") to stop it all being completely optimised away. That might show up a difference in the 'out_of_line' test where 15% on top on the data copies is massive - it may be that the data cache behaviour is very different for the two cases.
I tried using the following as the load:
volatile unsigned long foo;
static __always_inline size_t idle_user_iter(void __user *iter_from, size_t progress, size_t len, void *to, void *priv2) { nop(); nop(); foo += (unsigned long)iter_from; foo += (unsigned long)len; foo += (unsigned long)to + progress; nop(); nop(); return 0; }
static __always_inline size_t idle_kernel_iter(void *iter_from, size_t progress, size_t len, void *to, void *priv2) { nop(); nop(); foo += (unsigned long)iter_from; foo += (unsigned long)len; foo += (unsigned long)to + progress; nop(); nop(); return 0; }
size_t iov_iter_idle(struct iov_iter *iter, size_t len, void *priv) { return iterate_and_advance(iter, len, priv, idle_user_iter, idle_kernel_iter); } EXPORT_SYMBOL(iov_iter_idle);
adding various things into a volatile variable to prevent the optimiser from discarding the calculations.
I get:
iov_kunit_benchmark_bvec: avg 395 uS, stddev 46 uS iov_kunit_benchmark_bvec: avg 397 uS, stddev 38 uS iov_kunit_benchmark_bvec: avg 411 uS, stddev 57 uS iov_kunit_benchmark_bvec_outofline: avg 781 uS, stddev 5 uS iov_kunit_benchmark_bvec_outofline: avg 781 uS, stddev 6 uS iov_kunit_benchmark_bvec_outofline: avg 781 uS, stddev 7 uS iov_kunit_benchmark_bvec_split: avg 3599 uS, stddev 737 uS iov_kunit_benchmark_bvec_split: avg 3664 uS, stddev 838 uS iov_kunit_benchmark_bvec_split: avg 3669 uS, stddev 875 uS iov_kunit_benchmark_iovec: avg 472 uS, stddev 17 uS iov_kunit_benchmark_iovec: avg 506 uS, stddev 59 uS iov_kunit_benchmark_iovec: avg 525 uS, stddev 14 uS iov_kunit_benchmark_kvec: avg 421 uS, stddev 73 uS iov_kunit_benchmark_kvec: avg 428 uS, stddev 68 uS iov_kunit_benchmark_kvec: avg 469 uS, stddev 75 uS iov_kunit_benchmark_ubuf: avg 1052 uS, stddev 6 uS iov_kunit_benchmark_ubuf: avg 1168 uS, stddev 8 uS iov_kunit_benchmark_ubuf: avg 1168 uS, stddev 9 uS iov_kunit_benchmark_xarray: avg 680 uS, stddev 11 uS iov_kunit_benchmark_xarray: avg 682 uS, stddev 20 uS iov_kunit_benchmark_xarray: avg 686 uS, stddev 46 uS iov_kunit_benchmark_xarray_outofline: avg 1340 uS, stddev 34 uS iov_kunit_benchmark_xarray_outofline: avg 1358 uS, stddev 12 uS iov_kunit_benchmark_xarray_outofline: avg 1358 uS, stddev 15 uS
where I made the iovec and kvec tests split their buffers into PAGE_SIZE segments and the ubuf test issue an iteration per PAGE_SIZE'd chunk. Splitting kvec into just 8 results in the iteration taking <1uS.
The bvec_split test is doing a kmalloc() per 256 pages inside of the loop, which is why that takes quite a long time.
David
David Laight David.Laight@ACULAB.COM wrote:
Isn't that going to be completely dominated by the cache fills from memory?
I'd have thought you'd need to use something with a lot of small fragments so that the iteration code dominates the copy.
Okay, if I switch it to using MAP_ANON for the big 256MiB buffer, switch all the benchmarking tests to use copy_from_iter() rather than copy_to_iter() and make the iovec benchmark use a separate iovec for each page, there's then a single page replicated across the mapping.
Given that, without my macro-to-inline-func patches applied, I see:
iov_kunit_benchmark_bvec: avg 3184 uS, stddev 16 uS iov_kunit_benchmark_bvec: avg 3189 uS, stddev 17 uS iov_kunit_benchmark_bvec: avg 3190 uS, stddev 16 uS iov_kunit_benchmark_bvec_outofline: avg 3731 uS, stddev 10 uS iov_kunit_benchmark_bvec_outofline: avg 3735 uS, stddev 10 uS iov_kunit_benchmark_bvec_outofline: avg 3738 uS, stddev 11 uS iov_kunit_benchmark_bvec_split: avg 3403 uS, stddev 10 uS iov_kunit_benchmark_bvec_split: avg 3405 uS, stddev 18 uS iov_kunit_benchmark_bvec_split: avg 3407 uS, stddev 29 uS iov_kunit_benchmark_iovec: avg 6616 uS, stddev 20 uS iov_kunit_benchmark_iovec: avg 6619 uS, stddev 22 uS iov_kunit_benchmark_iovec: avg 6621 uS, stddev 46 uS iov_kunit_benchmark_kvec: avg 2671 uS, stddev 12 uS iov_kunit_benchmark_kvec: avg 2671 uS, stddev 13 uS iov_kunit_benchmark_kvec: avg 2675 uS, stddev 12 uS iov_kunit_benchmark_ubuf: avg 6191 uS, stddev 1946 uS iov_kunit_benchmark_ubuf: avg 6418 uS, stddev 3263 uS iov_kunit_benchmark_ubuf: avg 6443 uS, stddev 3275 uS iov_kunit_benchmark_xarray: avg 3689 uS, stddev 5 uS iov_kunit_benchmark_xarray: avg 3689 uS, stddev 6 uS iov_kunit_benchmark_xarray: avg 3698 uS, stddev 22 uS iov_kunit_benchmark_xarray_outofline: avg 4202 uS, stddev 3 uS iov_kunit_benchmark_xarray_outofline: avg 4204 uS, stddev 9 uS iov_kunit_benchmark_xarray_outofline: avg 4210 uS, stddev 9 uS
and with, I get:
iov_kunit_benchmark_bvec: avg 3241 uS, stddev 13 uS iov_kunit_benchmark_bvec: avg 3245 uS, stddev 16 uS iov_kunit_benchmark_bvec: avg 3248 uS, stddev 15 uS iov_kunit_benchmark_bvec_outofline: avg 3705 uS, stddev 12 uS iov_kunit_benchmark_bvec_outofline: avg 3706 uS, stddev 10 uS iov_kunit_benchmark_bvec_outofline: avg 3709 uS, stddev 9 uS iov_kunit_benchmark_bvec_split: avg 3446 uS, stddev 10 uS iov_kunit_benchmark_bvec_split: avg 3447 uS, stddev 12 uS iov_kunit_benchmark_bvec_split: avg 3448 uS, stddev 12 uS iov_kunit_benchmark_iovec: avg 6587 uS, stddev 22 uS iov_kunit_benchmark_iovec: avg 6587 uS, stddev 22 uS iov_kunit_benchmark_iovec: avg 6590 uS, stddev 27 uS iov_kunit_benchmark_kvec: avg 2671 uS, stddev 12 uS iov_kunit_benchmark_kvec: avg 2672 uS, stddev 12 uS iov_kunit_benchmark_kvec: avg 2676 uS, stddev 19 uS iov_kunit_benchmark_ubuf: avg 6241 uS, stddev 2199 uS iov_kunit_benchmark_ubuf: avg 6266 uS, stddev 2245 uS iov_kunit_benchmark_ubuf: avg 6513 uS, stddev 3899 uS iov_kunit_benchmark_xarray: avg 3695 uS, stddev 6 uS iov_kunit_benchmark_xarray: avg 3695 uS, stddev 7 uS iov_kunit_benchmark_xarray: avg 3703 uS, stddev 11 uS iov_kunit_benchmark_xarray_outofline: avg 4215 uS, stddev 16 uS iov_kunit_benchmark_xarray_outofline: avg 4217 uS, stddev 20 uS iov_kunit_benchmark_xarray_outofline: avg 4224 uS, stddev 10 uS
Interestingly, most of them are quite tight, but UBUF is all over the place. That's with the test covering the entire 256M span with a single UBUF iterator, so it would seem unlikely that the difference is due to the iteration framework.
David
linux-kselftest-mirror@lists.linaro.org
-
David Howells -
David Laight