diff options
Diffstat (limited to 'lib')
-rw-r--r-- | lib/Kconfig.debug | 11 | ||||
-rw-r--r-- | lib/Makefile | 1 | ||||
-rw-r--r-- | lib/idr.c | 2 | ||||
-rw-r--r-- | lib/iov_iter.c | 30 | ||||
-rw-r--r-- | lib/kunit/executor.c | 48 | ||||
-rw-r--r-- | lib/kunit/executor_test.c | 13 | ||||
-rw-r--r-- | lib/kunit/test.c | 3 | ||||
-rw-r--r-- | lib/kunit_iov_iter.c | 777 | ||||
-rw-r--r-- | lib/raid6/Makefile | 1 | ||||
-rw-r--r-- | lib/raid6/algos.c | 16 | ||||
-rw-r--r-- | lib/raid6/loongarch.h | 38 | ||||
-rw-r--r-- | lib/raid6/loongarch_simd.c | 422 | ||||
-rw-r--r-- | lib/raid6/recov_loongarch_simd.c | 513 | ||||
-rw-r--r-- | lib/raid6/test/Makefile | 12 | ||||
-rw-r--r-- | lib/xarray.c | 8 |
15 files changed, 1857 insertions, 38 deletions
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug index 319cfbeb0738..fa307f93fa2e 100644 --- a/lib/Kconfig.debug +++ b/lib/Kconfig.debug @@ -2237,6 +2237,17 @@ config TEST_DIV64 If unsure, say N. +config TEST_IOV_ITER + tristate "Test iov_iter operation" if !KUNIT_ALL_TESTS + depends on KUNIT + default KUNIT_ALL_TESTS + help + Enable this to turn on testing of the operation of the I/O iterator + (iov_iter). This test is executed only once during system boot (so + affects only boot time), or at module load time. + + If unsure, say N. + config KPROBES_SANITY_TEST tristate "Kprobes sanity tests" if !KUNIT_ALL_TESTS depends on DEBUG_KERNEL diff --git a/lib/Makefile b/lib/Makefile index 2e08397f6210..740109b6e2c8 100644 --- a/lib/Makefile +++ b/lib/Makefile @@ -64,6 +64,7 @@ obj-$(CONFIG_TEST_BITOPS) += test_bitops.o CFLAGS_test_bitops.o += -Werror obj-$(CONFIG_CPUMASK_KUNIT_TEST) += cpumask_kunit.o obj-$(CONFIG_TEST_SYSCTL) += test_sysctl.o +obj-$(CONFIG_TEST_IOV_ITER) += kunit_iov_iter.o obj-$(CONFIG_HASH_KUNIT_TEST) += test_hash.o obj-$(CONFIG_TEST_IDA) += test_ida.o obj-$(CONFIG_TEST_UBSAN) += test_ubsan.o diff --git a/lib/idr.c b/lib/idr.c index 7ecdfdb5309e..13f2758c2377 100644 --- a/lib/idr.c +++ b/lib/idr.c @@ -100,7 +100,7 @@ EXPORT_SYMBOL_GPL(idr_alloc); * @end: The maximum ID (exclusive). * @gfp: Memory allocation flags. * - * Allocates an unused ID in the range specified by @nextid and @end. If + * Allocates an unused ID in the range specified by @start and @end. If * @end is <= 0, it is treated as one larger than %INT_MAX. This allows * callers to use @start + N as @end as long as N is within integer range. * The search for an unused ID will start at the last ID allocated and will diff --git a/lib/iov_iter.c b/lib/iov_iter.c index b31597b0ca20..27234a820eeb 100644 --- a/lib/iov_iter.c +++ b/lib/iov_iter.c @@ -1654,14 +1654,14 @@ static ssize_t iov_iter_extract_bvec_pages(struct iov_iter *i, size_t *offset0) { struct page **p, *page; - size_t skip = i->iov_offset, offset; + size_t skip = i->iov_offset, offset, size; int k; for (;;) { if (i->nr_segs == 0) return 0; - maxsize = min(maxsize, i->bvec->bv_len - skip); - if (maxsize) + size = min(maxsize, i->bvec->bv_len - skip); + if (size) break; i->iov_offset = 0; i->nr_segs--; @@ -1674,16 +1674,16 @@ static ssize_t iov_iter_extract_bvec_pages(struct iov_iter *i, offset = skip % PAGE_SIZE; *offset0 = offset; - maxpages = want_pages_array(pages, maxsize, offset, maxpages); + maxpages = want_pages_array(pages, size, offset, maxpages); if (!maxpages) return -ENOMEM; p = *pages; for (k = 0; k < maxpages; k++) p[k] = page + k; - maxsize = min_t(size_t, maxsize, maxpages * PAGE_SIZE - offset); - iov_iter_advance(i, maxsize); - return maxsize; + size = min_t(size_t, size, maxpages * PAGE_SIZE - offset); + iov_iter_advance(i, size); + return size; } /* @@ -1698,14 +1698,14 @@ static ssize_t iov_iter_extract_kvec_pages(struct iov_iter *i, { struct page **p, *page; const void *kaddr; - size_t skip = i->iov_offset, offset, len; + size_t skip = i->iov_offset, offset, len, size; int k; for (;;) { if (i->nr_segs == 0) return 0; - maxsize = min(maxsize, i->kvec->iov_len - skip); - if (maxsize) + size = min(maxsize, i->kvec->iov_len - skip); + if (size) break; i->iov_offset = 0; i->nr_segs--; @@ -1717,13 +1717,13 @@ static ssize_t iov_iter_extract_kvec_pages(struct iov_iter *i, offset = (unsigned long)kaddr & ~PAGE_MASK; *offset0 = offset; - maxpages = want_pages_array(pages, maxsize, offset, maxpages); + maxpages = want_pages_array(pages, size, offset, maxpages); if (!maxpages) return -ENOMEM; p = *pages; kaddr -= offset; - len = offset + maxsize; + len = offset + size; for (k = 0; k < maxpages; k++) { size_t seg = min_t(size_t, len, PAGE_SIZE); @@ -1737,9 +1737,9 @@ static ssize_t iov_iter_extract_kvec_pages(struct iov_iter *i, kaddr += PAGE_SIZE; } - maxsize = min_t(size_t, maxsize, maxpages * PAGE_SIZE - offset); - iov_iter_advance(i, maxsize); - return maxsize; + size = min_t(size_t, size, maxpages * PAGE_SIZE - offset); + iov_iter_advance(i, size); + return size; } /* diff --git a/lib/kunit/executor.c b/lib/kunit/executor.c index 5181aa2e760b..a6348489d45f 100644 --- a/lib/kunit/executor.c +++ b/lib/kunit/executor.c @@ -65,7 +65,7 @@ struct kunit_glob_filter { }; /* Split "suite_glob.test_glob" into two. Assumes filter_glob is not empty. */ -static void kunit_parse_glob_filter(struct kunit_glob_filter *parsed, +static int kunit_parse_glob_filter(struct kunit_glob_filter *parsed, const char *filter_glob) { const int len = strlen(filter_glob); @@ -73,16 +73,28 @@ static void kunit_parse_glob_filter(struct kunit_glob_filter *parsed, if (!period) { parsed->suite_glob = kzalloc(len + 1, GFP_KERNEL); + if (!parsed->suite_glob) + return -ENOMEM; + parsed->test_glob = NULL; strcpy(parsed->suite_glob, filter_glob); - return; + return 0; } parsed->suite_glob = kzalloc(period - filter_glob + 1, GFP_KERNEL); + if (!parsed->suite_glob) + return -ENOMEM; + parsed->test_glob = kzalloc(len - (period - filter_glob) + 1, GFP_KERNEL); + if (!parsed->test_glob) { + kfree(parsed->suite_glob); + return -ENOMEM; + } strncpy(parsed->suite_glob, filter_glob, period - filter_glob); strncpy(parsed->test_glob, period + 1, len - (period - filter_glob)); + + return 0; } /* Create a copy of suite with only tests that match test_glob. */ @@ -152,21 +164,24 @@ kunit_filter_suites(const struct kunit_suite_set *suite_set, } copy_start = copy; - if (filter_glob) - kunit_parse_glob_filter(&parsed_glob, filter_glob); + if (filter_glob) { + *err = kunit_parse_glob_filter(&parsed_glob, filter_glob); + if (*err) + goto free_copy; + } /* Parse attribute filters */ if (filters) { filter_count = kunit_get_filter_count(filters); parsed_filters = kcalloc(filter_count, sizeof(*parsed_filters), GFP_KERNEL); if (!parsed_filters) { - kfree(copy); - return filtered; + *err = -ENOMEM; + goto free_parsed_glob; } for (j = 0; j < filter_count; j++) parsed_filters[j] = kunit_next_attr_filter(&filters, err); if (*err) - goto err; + goto free_parsed_filters; } for (i = 0; &suite_set->start[i] != suite_set->end; i++) { @@ -178,7 +193,7 @@ kunit_filter_suites(const struct kunit_suite_set *suite_set, parsed_glob.test_glob); if (IS_ERR(filtered_suite)) { *err = PTR_ERR(filtered_suite); - goto err; + goto free_parsed_filters; } } if (filter_count > 0 && parsed_filters != NULL) { @@ -195,10 +210,11 @@ kunit_filter_suites(const struct kunit_suite_set *suite_set, filtered_suite = new_filtered_suite; if (*err) - goto err; + goto free_parsed_filters; + if (IS_ERR(filtered_suite)) { *err = PTR_ERR(filtered_suite); - goto err; + goto free_parsed_filters; } if (!filtered_suite) break; @@ -213,17 +229,19 @@ kunit_filter_suites(const struct kunit_suite_set *suite_set, filtered.start = copy_start; filtered.end = copy; -err: - if (*err) - kfree(copy); +free_parsed_filters: + if (filter_count) + kfree(parsed_filters); +free_parsed_glob: if (filter_glob) { kfree(parsed_glob.suite_glob); kfree(parsed_glob.test_glob); } - if (filter_count) - kfree(parsed_filters); +free_copy: + if (*err) + kfree(copy); return filtered; } diff --git a/lib/kunit/executor_test.c b/lib/kunit/executor_test.c index 4084071d0eb5..b4f6f96b2844 100644 --- a/lib/kunit/executor_test.c +++ b/lib/kunit/executor_test.c @@ -119,7 +119,7 @@ static void parse_filter_attr_test(struct kunit *test) { int j, filter_count; struct kunit_attr_filter *parsed_filters; - char *filters = "speed>slow, module!=example"; + char filters[] = "speed>slow, module!=example", *filter = filters; int err = 0; filter_count = kunit_get_filter_count(filters); @@ -128,7 +128,7 @@ static void parse_filter_attr_test(struct kunit *test) parsed_filters = kunit_kcalloc(test, filter_count, sizeof(*parsed_filters), GFP_KERNEL); for (j = 0; j < filter_count; j++) { - parsed_filters[j] = kunit_next_attr_filter(&filters, &err); + parsed_filters[j] = kunit_next_attr_filter(&filter, &err); KUNIT_ASSERT_EQ_MSG(test, err, 0, "failed to parse filter '%s'", filters[j]); } @@ -154,6 +154,7 @@ static void filter_attr_test(struct kunit *test) .start = subsuite, .end = &subsuite[2], }; struct kunit_suite_set got; + char filter[] = "speed>slow"; int err = 0; subsuite[0] = alloc_fake_suite(test, "normal_suite", dummy_attr_test_cases); @@ -168,7 +169,7 @@ static void filter_attr_test(struct kunit *test) * attribute is unset and thus, the filtering is based on the parent attribute * of slow. */ - got = kunit_filter_suites(&suite_set, NULL, "speed>slow", NULL, &err); + got = kunit_filter_suites(&suite_set, NULL, filter, NULL, &err); KUNIT_ASSERT_NOT_ERR_OR_NULL(test, got.start); KUNIT_ASSERT_EQ(test, err, 0); kfree_at_end(test, got.start); @@ -191,12 +192,13 @@ static void filter_attr_empty_test(struct kunit *test) .start = subsuite, .end = &subsuite[2], }; struct kunit_suite_set got; + char filter[] = "module!=dummy"; int err = 0; subsuite[0] = alloc_fake_suite(test, "suite1", dummy_attr_test_cases); subsuite[1] = alloc_fake_suite(test, "suite2", dummy_attr_test_cases); - got = kunit_filter_suites(&suite_set, NULL, "module!=dummy", NULL, &err); + got = kunit_filter_suites(&suite_set, NULL, filter, NULL, &err); KUNIT_ASSERT_EQ(test, err, 0); kfree_at_end(test, got.start); /* just in case */ @@ -211,12 +213,13 @@ static void filter_attr_skip_test(struct kunit *test) .start = subsuite, .end = &subsuite[1], }; struct kunit_suite_set got; + char filter[] = "speed>slow"; int err = 0; subsuite[0] = alloc_fake_suite(test, "suite", dummy_attr_test_cases); /* Want: suite(slow, normal), NULL -> suite(slow with SKIP, normal), NULL */ - got = kunit_filter_suites(&suite_set, NULL, "speed>slow", "skip", &err); + got = kunit_filter_suites(&suite_set, NULL, filter, "skip", &err); KUNIT_ASSERT_NOT_ERR_OR_NULL(test, got.start); KUNIT_ASSERT_EQ(test, err, 0); kfree_at_end(test, got.start); diff --git a/lib/kunit/test.c b/lib/kunit/test.c index 49698a168437..421f13981412 100644 --- a/lib/kunit/test.c +++ b/lib/kunit/test.c @@ -784,12 +784,13 @@ static int kunit_module_notify(struct notifier_block *nb, unsigned long val, switch (val) { case MODULE_STATE_LIVE: - kunit_module_init(mod); break; case MODULE_STATE_GOING: kunit_module_exit(mod); break; case MODULE_STATE_COMING: + kunit_module_init(mod); + break; case MODULE_STATE_UNFORMED: break; } diff --git a/lib/kunit_iov_iter.c b/lib/kunit_iov_iter.c new file mode 100644 index 000000000000..859b67c4d697 --- /dev/null +++ b/lib/kunit_iov_iter.c @@ -0,0 +1,777 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* I/O iterator tests. This can only test kernel-backed iterator types. + * + * Copyright (C) 2023 Red Hat, Inc. All Rights Reserved. + * Written by David Howells (dhowells@redhat.com) + */ + +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt + +#include <linux/module.h> +#include <linux/vmalloc.h> +#include <linux/mm.h> +#include <linux/uio.h> +#include <linux/bvec.h> +#include <kunit/test.h> + +MODULE_DESCRIPTION("iov_iter testing"); +MODULE_AUTHOR("David Howells <dhowells@redhat.com>"); +MODULE_LICENSE("GPL"); + +struct kvec_test_range { + int 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 }, + { -1 } +}; + +static inline u8 pattern(unsigned long x) +{ + return x & 0xff; +} + +static void iov_kunit_unmap(void *data) +{ + vunmap(data); +} + +static void *__init iov_kunit_create_buffer(struct kunit *test, + struct page ***ppages, + size_t npages) +{ + struct page **pages; + unsigned long got; + void *buffer; + + pages = kunit_kcalloc(test, npages, sizeof(struct page *), GFP_KERNEL); + KUNIT_ASSERT_NOT_ERR_OR_NULL(test, pages); + *ppages = pages; + + got = alloc_pages_bulk_array(GFP_KERNEL, npages, pages); + if (got != npages) { + release_pages(pages, got); + KUNIT_ASSERT_EQ(test, got, npages); + } + + buffer = vmap(pages, npages, VM_MAP | VM_MAP_PUT_PAGES, PAGE_KERNEL); + KUNIT_ASSERT_NOT_ERR_OR_NULL(test, buffer); + + kunit_add_action_or_reset(test, iov_kunit_unmap, 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, + void *buffer, size_t bufsize, + const struct kvec_test_range *pr) +{ + size_t size = 0; + int i; + + for (i = 0; i < kvmax; i++, pr++) { + if (pr->from < 0) + break; + KUNIT_ASSERT_GE(test, pr->to, pr->from); + KUNIT_ASSERT_LE(test, pr->to, bufsize); + kvec[i].iov_base = buffer + pr->from; + kvec[i].iov_len = pr->to - pr->from; + size += pr->to - pr->from; + } + KUNIT_ASSERT_LE(test, size, bufsize); + + iov_iter_kvec(iter, dir, kvec, i, size); +} + +/* + * Test copying to a ITER_KVEC-type iterator. + */ +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; + + 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_buffer(test, &bpages, npages); + memset(buffer, 0, bufsize); + + iov_kunit_load_kvec(test, &iter, READ, kvec, ARRAY_SIZE(kvec), + 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); + + /* 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; + } + + KUNIT_SUCCEED(); +} + +/* + * Test copying from a ITER_KVEC-type iterator. + */ +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; + + bufsize = 0x100000; + npages = bufsize / PAGE_SIZE; + + buffer = iov_kunit_create_buffer(test, &bpages, npages); + for (i = 0; i < bufsize; i++) + buffer[i] = pattern(i); + + scratch = iov_kunit_create_buffer(test, &spages, npages); + memset(scratch, 0, bufsize); + + iov_kunit_load_kvec(test, &iter, WRITE, kvec, ARRAY_SIZE(kvec), + buffer, bufsize, kvec_test_ranges); + size = min(iter.count, bufsize); + + 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); + + /* 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; + } + + 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, -1, -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) +{ + struct page *can_merge = NULL, *page; + size_t size = 0; + int i; + + for (i = 0; i < bvmax; i++, pr++) { + if (pr->from < 0) + break; + KUNIT_ASSERT_LT(test, pr->page, npages); + KUNIT_ASSERT_LT(test, pr->page * PAGE_SIZE, bufsize); + KUNIT_ASSERT_GE(test, pr->from, 0); + KUNIT_ASSERT_GE(test, pr->to, pr->from); + KUNIT_ASSERT_LE(test, pr->to, PAGE_SIZE); + + page = pages[pr->page]; + if (pr->from == 0 && pr->from != pr->to && page == can_merge) { + i--; + bvec[i].bv_len += pr->to; + } else { + bvec_set_page(&bvec[i], page, pr->to - pr->from, pr->from); + } + + size += pr->to - pr->from; + if ((pr->to & ~PAGE_MASK) == 0) + can_merge = page + pr->to / PAGE_SIZE; + else + can_merge = NULL; + } + + iov_iter_bvec(iter, dir, bvec, i, size); +} + +/* + * Test copying to a ITER_BVEC-type iterator. + */ +static void __init iov_kunit_copy_to_bvec(struct kunit *test) +{ + const struct bvec_test_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; + + 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_buffer(test, &bpages, npages); + memset(buffer, 0, bufsize); + + iov_kunit_load_bvec(test, &iter, READ, bvec, ARRAY_SIZE(bvec), + bpages, npages, bufsize, bvec_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); + + /* Build the expected image in the scratch buffer. */ + b = 0; + patt = 0; + memset(scratch, 0, bufsize); + for (pr = bvec_test_ranges; pr->from >= 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; + } + + KUNIT_SUCCEED(); +} + +/* + * Test copying from a ITER_BVEC-type iterator. + */ +static void __init iov_kunit_copy_from_bvec(struct kunit *test) +{ + const struct bvec_test_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; + + bufsize = 0x100000; + npages = bufsize / PAGE_SIZE; + + buffer = iov_kunit_create_buffer(test, &bpages, npages); + for (i = 0; i < bufsize; i++) + buffer[i] = pattern(i); + + scratch = iov_kunit_create_buffer(test, &spages, npages); + memset(scratch, 0, bufsize); + + iov_kunit_load_bvec(test, &iter, WRITE, bvec, ARRAY_SIZE(bvec), + bpages, npages, bufsize, bvec_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); + + /* Build the expected image in the main buffer. */ + i = 0; + memset(buffer, 0, bufsize); + for (pr = bvec_test_ranges; pr->from >= 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: + + /* 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; + } + + KUNIT_SUCCEED(); +} + +static void iov_kunit_destroy_xarray(void *data) +{ + struct xarray *xarray = data; + + xa_destroy(xarray); + kfree(xarray); +} + +static void __init iov_kunit_load_xarray(struct kunit *test, + struct iov_iter *iter, int dir, + struct xarray *xarray, + struct page **pages, size_t npages) +{ + size_t size = 0; + int i; + + for (i = 0; i < npages; i++) { + void *x = xa_store(xarray, i, pages[i], GFP_KERNEL); + + KUNIT_ASSERT_FALSE(test, xa_is_err(x)); + size += PAGE_SIZE; + } + iov_iter_xarray(iter, dir, xarray, 0, size); +} + +static struct xarray *iov_kunit_create_xarray(struct kunit *test) +{ + struct xarray *xarray; + + xarray = kzalloc(sizeof(struct xarray), GFP_KERNEL); + xa_init(xarray); + KUNIT_ASSERT_NOT_ERR_OR_NULL(test, xarray); + kunit_add_action_or_reset(test, iov_kunit_destroy_xarray, xarray); + return xarray; +} + +/* + * Test copying to a ITER_XARRAY-type iterator. + */ +static void __init iov_kunit_copy_to_xarray(struct kunit *test) +{ + const struct kvec_test_range *pr; + struct iov_iter iter; + struct xarray *xarray; + struct page **spages, **bpages; + u8 *scratch, *buffer; + size_t bufsize, npages, size, copied; + int i, patt; + + bufsize = 0x100000; + npages = bufsize / PAGE_SIZE; + + xarray = iov_kunit_create_xarray(test); + + scratch = iov_kunit_create_buffer(test, &spages, npages); + for (i = 0; i < bufsize; i++) + scratch[i] = pattern(i); + + buffer = iov_kunit_create_buffer(test, &bpages, npages); + memset(buffer, 0, bufsize); + + iov_kunit_load_xarray(test, &iter, READ, xarray, bpages, npages); + + i = 0; + for (pr = kvec_test_ranges; pr->from >= 0; pr++) { + size = pr->to - pr->from; + KUNIT_ASSERT_LE(test, pr->to, bufsize); + + iov_iter_xarray(&iter, READ, xarray, 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); + 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; + } + + KUNIT_SUCCEED(); +} + +/* + * Test copying from a ITER_XARRAY-type iterator. + */ +static void __init iov_kunit_copy_from_xarray(struct kunit *test) +{ + const struct kvec_test_range *pr; + struct iov_iter iter; + struct xarray *xarray; + struct page **spages, **bpages; + u8 *scratch, *buffer; + size_t bufsize, npages, size, copied; + int i, j; + + bufsize = 0x100000; + npages = bufsize / PAGE_SIZE; + + xarray = iov_kunit_create_xarray(test); + + buffer = iov_kunit_create_buffer(test, &bpages, npages); + for (i = 0; i < bufsize; i++) + buffer[i] = pattern(i); + + scratch = iov_kunit_create_buffer(test, &spages, npages); + memset(scratch, 0, bufsize); + + iov_kunit_load_xarray(test, &iter, READ, xarray, bpages, npages); + + i = 0; + for (pr = kvec_test_ranges; pr->from >= 0; pr++) { + size = pr->to - pr->from; + KUNIT_ASSERT_LE(test, pr->to, bufsize); + + iov_iter_xarray(&iter, WRITE, xarray, 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); + 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; + } + + KUNIT_SUCCEED(); +} + +/* + * Test the extraction of ITER_KVEC-type iterators. + */ +static void __init iov_kunit_extract_pages_kvec(struct kunit *test) +{ + const struct kvec_test_range *pr; + struct iov_iter iter; + struct page **bpages, *pagelist[8], **pages = pagelist; + struct kvec kvec[8]; + u8 *buffer; + ssize_t len; + size_t bufsize, size = 0, npages; + int i, from; + + bufsize = 0x100000; + npages = bufsize / PAGE_SIZE; + + buffer = iov_kunit_create_buffer(test, &bpages, npages); + + iov_kunit_load_kvec(test, &iter, READ, kvec, ARRAY_SIZE(kvec), + 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 *)(unsigned long)0xaa55aa55aa55aa55ULL; + + 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_GE(test, (ssize_t)offset0, 0); + KUNIT_EXPECT_LT(test, offset0, PAGE_SIZE); + KUNIT_EXPECT_LE(test, len, size); + KUNIT_EXPECT_EQ(test, iter.count, size - len); + size -= len; + + if (len == 0) + break; + + 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 (from < 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_BVEC-type iterators. + */ +static void __init iov_kunit_extract_pages_bvec(struct kunit *test) +{ + const struct bvec_test_range *pr; + struct iov_iter iter; + struct page **bpages, *pagelist[8], **pages = pagelist; + struct bio_vec bvec[8]; + ssize_t len; + size_t bufsize, size = 0, npages; + int i, from; + + bufsize = 0x100000; + npages = bufsize / PAGE_SIZE; + + iov_kunit_create_buffer(test, &bpages, npages); + iov_kunit_load_bvec(test, &iter, READ, bvec, ARRAY_SIZE(bvec), + bpages, npages, bufsize, bvec_test_ranges); + size = iter.count; + + pr = bvec_test_ranges; + from = pr->from; + do { + size_t offset0 = LONG_MAX; + + for (i = 0; i < ARRAY_SIZE(pagelist); i++) + pagelist[i] = (void *)(unsigned long)0xaa55aa55aa55aa55ULL; + + 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_GE(test, (ssize_t)offset0, 0); + KUNIT_EXPECT_LT(test, offset0, PAGE_SIZE); + KUNIT_EXPECT_LE(test, len, size); + KUNIT_EXPECT_EQ(test, iter.count, size - len); + size -= len; + + if (len == 0) + break; + + 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 (from < 0) + goto stop; + } + ix = pr->page + 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_XARRAY-type iterators. + */ +static void __init iov_kunit_extract_pages_xarray(struct kunit *test) +{ + const struct kvec_test_range *pr; + struct iov_iter iter; + struct xarray *xarray; + struct page **bpages, *pagelist[8], **pages = pagelist; + ssize_t len; + size_t bufsize, size = 0, npages; + int i, from; + + bufsize = 0x100000; + npages = bufsize / PAGE_SIZE; + + xarray = iov_kunit_create_xarray(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++) { + from = pr->from; + size = pr->to - from; + KUNIT_ASSERT_LE(test, pr->to, bufsize); + + iov_iter_xarray(&iter, WRITE, xarray, from, size); + + do { + size_t offset0 = LONG_MAX; + + for (i = 0; i < ARRAY_SIZE(pagelist); i++) + pagelist[i] = (void *)(unsigned long)0xaa55aa55aa55aa55ULL; + + 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); + + 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(); +} + +static struct kunit_case __refdata iov_kunit_cases[] = { + KUNIT_CASE(iov_kunit_copy_to_kvec), + KUNIT_CASE(iov_kunit_copy_from_kvec), + KUNIT_CASE(iov_kunit_copy_to_bvec), + 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_kvec), + KUNIT_CASE(iov_kunit_extract_pages_bvec), + KUNIT_CASE(iov_kunit_extract_pages_xarray), + {} +}; + +static struct kunit_suite iov_kunit_suite = { + .name = "iov_iter", + .test_cases = iov_kunit_cases, +}; + +kunit_test_suites(&iov_kunit_suite); diff --git a/lib/raid6/Makefile b/lib/raid6/Makefile index 45e17619422b..035b0a4db476 100644 --- a/lib/raid6/Makefile +++ b/lib/raid6/Makefile @@ -9,6 +9,7 @@ raid6_pq-$(CONFIG_ALTIVEC) += altivec1.o altivec2.o altivec4.o altivec8.o \ vpermxor1.o vpermxor2.o vpermxor4.o vpermxor8.o raid6_pq-$(CONFIG_KERNEL_MODE_NEON) += neon.o neon1.o neon2.o neon4.o neon8.o recov_neon.o recov_neon_inner.o raid6_pq-$(CONFIG_S390) += s390vx8.o recov_s390xc.o +raid6_pq-$(CONFIG_LOONGARCH) += loongarch_simd.o recov_loongarch_simd.o hostprogs += mktables diff --git a/lib/raid6/algos.c b/lib/raid6/algos.c index a22a05c9af8a..0ec534faf019 100644 --- a/lib/raid6/algos.c +++ b/lib/raid6/algos.c @@ -73,6 +73,14 @@ const struct raid6_calls * const raid6_algos[] = { &raid6_neonx2, &raid6_neonx1, #endif +#ifdef CONFIG_LOONGARCH +#ifdef CONFIG_CPU_HAS_LASX + &raid6_lasx, +#endif +#ifdef CONFIG_CPU_HAS_LSX + &raid6_lsx, +#endif +#endif #if defined(__ia64__) &raid6_intx32, &raid6_intx16, @@ -104,6 +112,14 @@ const struct raid6_recov_calls *const raid6_recov_algos[] = { #if defined(CONFIG_KERNEL_MODE_NEON) &raid6_recov_neon, #endif +#ifdef CONFIG_LOONGARCH +#ifdef CONFIG_CPU_HAS_LASX + &raid6_recov_lasx, +#endif +#ifdef CONFIG_CPU_HAS_LSX + &raid6_recov_lsx, +#endif +#endif &raid6_recov_intx1, NULL }; diff --git a/lib/raid6/loongarch.h b/lib/raid6/loongarch.h new file mode 100644 index 000000000000..acfc33ce7056 --- /dev/null +++ b/lib/raid6/loongarch.h @@ -0,0 +1,38 @@ +/* SPDX-License-Identifier: GPL-2.0-or-later */ +/* + * Copyright (C) 2023 WANG Xuerui <git@xen0n.name> + * + * raid6/loongarch.h + * + * Definitions common to LoongArch RAID-6 code only + */ + +#ifndef _LIB_RAID6_LOONGARCH_H +#define _LIB_RAID6_LOONGARCH_H + +#ifdef __KERNEL__ + +#include <asm/cpu-features.h> +#include <asm/fpu.h> + +#else /* for user-space testing */ + +#include <sys/auxv.h> + +/* have to supply these defines for glibc 2.37- and musl */ +#ifndef HWCAP_LOONGARCH_LSX +#define HWCAP_LOONGARCH_LSX (1 << 4) +#endif +#ifndef HWCAP_LOONGARCH_LASX +#define HWCAP_LOONGARCH_LASX (1 << 5) +#endif + +#define kernel_fpu_begin() +#define kernel_fpu_end() + +#define cpu_has_lsx (getauxval(AT_HWCAP) & HWCAP_LOONGARCH_LSX) +#define cpu_has_lasx (getauxval(AT_HWCAP) & HWCAP_LOONGARCH_LASX) + +#endif /* __KERNEL__ */ + +#endif /* _LIB_RAID6_LOONGARCH_H */ diff --git a/lib/raid6/loongarch_simd.c b/lib/raid6/loongarch_simd.c new file mode 100644 index 000000000000..aa5d9f924ca3 --- /dev/null +++ b/lib/raid6/loongarch_simd.c @@ -0,0 +1,422 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* + * RAID6 syndrome calculations in LoongArch SIMD (LSX & LASX) + * + * Copyright 2023 WANG Xuerui <git@xen0n.name> + * + * Based on the generic RAID-6 code (int.uc): + * + * Copyright 2002-2004 H. Peter Anvin + */ + +#include <linux/raid/pq.h> +#include "loongarch.h" + +/* + * The vector algorithms are currently priority 0, which means the generic + * scalar algorithms are not being disabled if vector support is present. + * This is like the similar LoongArch RAID5 XOR code, with the main reason + * repeated here: it cannot be ruled out at this point of time, that some + * future (maybe reduced) models could run the vector algorithms slower than + * the scalar ones, maybe for errata or micro-op reasons. It may be + * appropriate to revisit this after one or two more uarch generations. + */ + +#ifdef CONFIG_CPU_HAS_LSX +#define NSIZE 16 + +static int raid6_has_lsx(void) +{ + return cpu_has_lsx; +} + +static void raid6_lsx_gen_syndrome(int disks, size_t bytes, void **ptrs) +{ + u8 **dptr = (u8 **)ptrs; + u8 *p, *q; + int d, z, z0; + + z0 = disks - 3; /* Highest data disk */ + p = dptr[z0+1]; /* XOR parity */ + q = dptr[z0+2]; /* RS syndrome */ + + kernel_fpu_begin(); + + /* + * $vr0, $vr1, $vr2, $vr3: wp + * $vr4, $vr5, $vr6, $vr7: wq + * $vr8, $vr9, $vr10, $vr11: wd + * $vr12, $vr13, $vr14, $vr15: w2 + * $vr16, $vr17, $vr18, $vr19: w1 + */ + for (d = 0; d < bytes; d += NSIZE*4) { + /* wq$$ = wp$$ = *(unative_t *)&dptr[z0][d+$$*NSIZE]; */ + asm volatile("vld $vr0, %0" : : "m"(dptr[z0][d+0*NSIZE])); + asm volatile("vld $vr1, %0" : : "m"(dptr[z0][d+1*NSIZE])); + asm volatile("vld $vr2, %0" : : "m"(dptr[z0][d+2*NSIZE])); + asm volatile("vld $vr3, %0" : : "m"(dptr[z0][d+3*NSIZE])); + asm volatile("vori.b $vr4, $vr0, 0"); + asm volatile("vori.b $vr5, $vr1, 0"); + asm volatile("vori.b $vr6, $vr2, 0"); + asm volatile("vori.b $vr7, $vr3, 0"); + for (z = z0-1; z >= 0; z--) { + /* wd$$ = *(unative_t *)&dptr[z][d+$$*NSIZE]; */ + asm volatile("vld $vr8, %0" : : "m"(dptr[z][d+0*NSIZE])); + asm volatile("vld $vr9, %0" : : "m"(dptr[z][d+1*NSIZE])); + asm volatile("vld $vr10, %0" : : "m"(dptr[z][d+2*NSIZE])); + asm volatile("vld $vr11, %0" : : "m"(dptr[z][d+3*NSIZE])); + /* wp$$ ^= wd$$; */ + asm volatile("vxor.v $vr0, $vr0, $vr8"); + asm volatile("vxor.v $vr1, $vr1, $vr9"); + asm volatile("vxor.v $vr2, $vr2, $vr10"); + asm volatile("vxor.v $vr3, $vr3, $vr11"); + /* w2$$ = MASK(wq$$); */ + asm volatile("vslti.b $vr12, $vr4, 0"); + asm volatile("vslti.b $vr13, $vr5, 0"); + asm volatile("vslti.b $vr14, $vr6, 0"); + asm volatile("vslti.b $vr15, $vr7, 0"); + /* w1$$ = SHLBYTE(wq$$); */ + asm volatile("vslli.b $vr16, $vr4, 1"); + asm volatile("vslli.b $vr17, $vr5, 1"); + asm volatile("vslli.b $vr18, $vr6, 1"); + asm volatile("vslli.b $vr19, $vr7, 1"); + /* w2$$ &= NBYTES(0x1d); */ + asm volatile("vandi.b $vr12, $vr12, 0x1d"); + asm volatile("vandi.b $vr13, $vr13, 0x1d"); + asm volatile("vandi.b $vr14, $vr14, 0x1d"); + asm volatile("vandi.b $vr15, $vr15, 0x1d"); + /* w1$$ ^= w2$$; */ + asm volatile("vxor.v $vr16, $vr16, $vr12"); + asm volatile("vxor.v $vr17, $vr17, $vr13"); + asm volatile("vxor.v $vr18, $vr18, $vr14"); + asm volatile("vxor.v $vr19, $vr19, $vr15"); + /* wq$$ = w1$$ ^ wd$$; */ + asm volatile("vxor.v $vr4, $vr16, $vr8"); + asm volatile("vxor.v $vr5, $vr17, $vr9"); + asm volatile("vxor.v $vr6, $vr18, $vr10"); + asm volatile("vxor.v $vr7, $vr19, $vr11"); + } + /* *(unative_t *)&p[d+NSIZE*$$] = wp$$; */ + asm volatile("vst $vr0, %0" : "=m"(p[d+NSIZE*0])); + asm volatile("vst $vr1, %0" : "=m"(p[d+NSIZE*1])); + asm volatile("vst $vr2, %0" : "=m"(p[d+NSIZE*2])); + asm volatile("vst $vr3, %0" : "=m"(p[d+NSIZE*3])); + /* *(unative_t *)&q[d+NSIZE*$$] = wq$$; */ + asm volatile("vst $vr4, %0" : "=m"(q[d+NSIZE*0])); + asm volatile("vst $vr5, %0" : "=m"(q[d+NSIZE*1])); + asm volatile("vst $vr6, %0" : "=m"(q[d+NSIZE*2])); + asm volatile("vst $vr7, %0" : "=m"(q[d+NSIZE*3])); + } + + kernel_fpu_end(); +} + +static void raid6_lsx_xor_syndrome(int disks, int start, int stop, + size_t bytes, void **ptrs) +{ + u8 **dptr = (u8 **)ptrs; + u8 *p, *q; + int d, z, z0; + + z0 = stop; /* P/Q right side optimization */ + p = dptr[disks-2]; /* XOR parity */ + q = dptr[disks-1]; /* RS syndrome */ + + kernel_fpu_begin(); + + /* + * $vr0, $vr1, $vr2, $vr3: wp + * $vr4, $vr5, $vr6, $vr7: wq + * $vr8, $vr9, $vr10, $vr11: wd + * $vr12, $vr13, $vr14, $vr15: w2 + * $vr16, $vr17, $vr18, $vr19: w1 + */ + for (d = 0; d < bytes; d += NSIZE*4) { + /* P/Q data pages */ + /* wq$$ = wp$$ = *(unative_t *)&dptr[z0][d+$$*NSIZE]; */ + asm volatile("vld $vr0, %0" : : "m"(dptr[z0][d+0*NSIZE])); + asm volatile("vld $vr1, %0" : : "m"(dptr[z0][d+1*NSIZE])); + asm volatile("vld $vr2, %0" : : "m"(dptr[z0][d+2*NSIZE])); + asm volatile("vld $vr3, %0" : : "m"(dptr[z0][d+3*NSIZE])); + asm volatile("vori.b $vr4, $vr0, 0"); + asm volatile("vori.b $vr5, $vr1, 0"); + asm volatile("vori.b $vr6, $vr2, 0"); + asm volatile("vori.b $vr7, $vr3, 0"); + for (z = z0-1; z >= start; z--) { + /* wd$$ = *(unative_t *)&dptr[z][d+$$*NSIZE]; */ + asm volatile("vld $vr8, %0" : : "m"(dptr[z][d+0*NSIZE])); + asm volatile("vld $vr9, %0" : : "m"(dptr[z][d+1*NSIZE])); + asm volatile("vld $vr10, %0" : : "m"(dptr[z][d+2*NSIZE])); + asm volatile("vld $vr11, %0" : : "m"(dptr[z][d+3*NSIZE])); + /* wp$$ ^= wd$$; */ + asm volatile("vxor.v $vr0, $vr0, $vr8"); + asm volatile("vxor.v $vr1, $vr1, $vr9"); + asm volatile("vxor.v $vr2, $vr2, $vr10"); + asm volatile("vxor.v $vr3, $vr3, $vr11"); + /* w2$$ = MASK(wq$$); */ + asm volatile("vslti.b $vr12, $vr4, 0"); + asm volatile("vslti.b $vr13, $vr5, 0"); + asm volatile("vslti.b $vr14, $vr6, 0"); + asm volatile("vslti.b $vr15, $vr7, 0"); + /* w1$$ = SHLBYTE(wq$$); */ + asm volatile("vslli.b $vr16, $vr4, 1"); + asm volatile("vslli.b $vr17, $vr5, 1"); + asm volatile("vslli.b $vr18, $vr6, 1"); + asm volatile("vslli.b $vr19, $vr7, 1"); + /* w2$$ &= NBYTES(0x1d); */ + asm volatile("vandi.b $vr12, $vr12, 0x1d"); + asm volatile("vandi.b $vr13, $vr13, 0x1d"); + asm volatile("vandi.b $vr14, $vr14, 0x1d"); + asm volatile("vandi.b $vr15, $vr15, 0x1d"); + /* w1$$ ^= w2$$; */ + asm volatile("vxor.v $vr16, $vr16, $vr12"); + asm volatile("vxor.v $vr17, $vr17, $vr13"); + asm volatile("vxor.v $vr18, $vr18, $vr14"); + asm volatile("vxor.v $vr19, $vr19, $vr15"); + /* wq$$ = w1$$ ^ wd$$; */ + asm volatile("vxor.v $vr4, $vr16, $vr8"); + asm volatile("vxor.v $vr5, $vr17, $vr9"); + asm volatile("vxor.v $vr6, $vr18, $vr10"); + asm volatile("vxor.v $vr7, $vr19, $vr11"); + } + + /* P/Q left side optimization */ + for (z = start-1; z >= 0; z--) { + /* w2$$ = MASK(wq$$); */ + asm volatile("vslti.b $vr12, $vr4, 0"); + asm volatile("vslti.b $vr13, $vr5, 0"); + asm volatile("vslti.b $vr14, $vr6, 0"); + asm volatile("vslti.b $vr15, $vr7, 0"); + /* w1$$ = SHLBYTE(wq$$); */ + asm volatile("vslli.b $vr16, $vr4, 1"); + asm volatile("vslli.b $vr17, $vr5, 1"); + asm volatile("vslli.b $vr18, $vr6, 1"); + asm volatile("vslli.b $vr19, $vr7, 1"); + /* w2$$ &= NBYTES(0x1d); */ + asm volatile("vandi.b $vr12, $vr12, 0x1d"); + asm volatile("vandi.b $vr13, $vr13, 0x1d"); + asm volatile("vandi.b $vr14, $vr14, 0x1d"); + asm volatile("vandi.b $vr15, $vr15, 0x1d"); + /* wq$$ = w1$$ ^ w2$$; */ + asm volatile("vxor.v $vr4, $vr16, $vr12"); + asm volatile("vxor.v $vr5, $vr17, $vr13"); + asm volatile("vxor.v $vr6, $vr18, $vr14"); + asm volatile("vxor.v $vr7, $vr19, $vr15"); + } + /* + * *(unative_t *)&p[d+NSIZE*$$] ^= wp$$; + * *(unative_t *)&q[d+NSIZE*$$] ^= wq$$; + */ + asm volatile( + "vld $vr20, %0\n\t" + "vld $vr21, %1\n\t" + "vld $vr22, %2\n\t" + "vld $vr23, %3\n\t" + "vld $vr24, %4\n\t" + "vld $vr25, %5\n\t" + "vld $vr26, %6\n\t" + "vld $vr27, %7\n\t" + "vxor.v $vr20, $vr20, $vr0\n\t" + "vxor.v $vr21, $vr21, $vr1\n\t" + "vxor.v $vr22, $vr22, $vr2\n\t" + "vxor.v $vr23, $vr23, $vr3\n\t" + "vxor.v $vr24, $vr24, $vr4\n\t" + "vxor.v $vr25, $vr25, $vr5\n\t" + "vxor.v $vr26, $vr26, $vr6\n\t" + "vxor.v $vr27, $vr27, $vr7\n\t" + "vst $vr20, %0\n\t" + "vst $vr21, %1\n\t" + "vst $vr22, %2\n\t" + "vst $vr23, %3\n\t" + "vst $vr24, %4\n\t" + "vst $vr25, %5\n\t" + "vst $vr26, %6\n\t" + "vst $vr27, %7\n\t" + : "+m"(p[d+NSIZE*0]), "+m"(p[d+NSIZE*1]), + "+m"(p[d+NSIZE*2]), "+m"(p[d+NSIZE*3]), + "+m"(q[d+NSIZE*0]), "+m"(q[d+NSIZE*1]), + "+m"(q[d+NSIZE*2]), "+m"(q[d+NSIZE*3]) + ); + } + + kernel_fpu_end(); +} + +const struct raid6_calls raid6_lsx = { + raid6_lsx_gen_syndrome, + raid6_lsx_xor_syndrome, + raid6_has_lsx, + "lsx", + .priority = 0 /* see the comment near the top of the file for reason */ +}; + +#undef NSIZE +#endif /* CONFIG_CPU_HAS_LSX */ + +#ifdef CONFIG_CPU_HAS_LASX +#define NSIZE 32 + +static int raid6_has_lasx(void) +{ + return cpu_has_lasx; +} + +static void raid6_lasx_gen_syndrome(int disks, size_t bytes, void **ptrs) +{ + u8 **dptr = (u8 **)ptrs; + u8 *p, *q; + int d, z, z0; + + z0 = disks - 3; /* Highest data disk */ + p = dptr[z0+1]; /* XOR parity */ + q = dptr[z0+2]; /* RS syndrome */ + + kernel_fpu_begin(); + + /* + * $xr0, $xr1: wp + * $xr2, $xr3: wq + * $xr4, $xr5: wd + * $xr6, $xr7: w2 + * $xr8, $xr9: w1 + */ + for (d = 0; d < bytes; d += NSIZE*2) { + /* wq$$ = wp$$ = *(unative_t *)&dptr[z0][d+$$*NSIZE]; */ + asm volatile("xvld $xr0, %0" : : "m"(dptr[z0][d+0*NSIZE])); + asm volatile("xvld $xr1, %0" : : "m"(dptr[z0][d+1*NSIZE])); + asm volatile("xvori.b $xr2, $xr0, 0"); + asm volatile("xvori.b $xr3, $xr1, 0"); + for (z = z0-1; z >= 0; z--) { + /* wd$$ = *(unative_t *)&dptr[z][d+$$*NSIZE]; */ + asm volatile("xvld $xr4, %0" : : "m"(dptr[z][d+0*NSIZE])); + asm volatile("xvld $xr5, %0" : : "m"(dptr[z][d+1*NSIZE])); + /* wp$$ ^= wd$$; */ + asm volatile("xvxor.v $xr0, $xr0, $xr4"); + asm volatile("xvxor.v $xr1, $xr1, $xr5"); + /* w2$$ = MASK(wq$$); */ + asm volatile("xvslti.b $xr6, $xr2, 0"); + asm volatile("xvslti.b $xr7, $xr3, 0"); + /* w1$$ = SHLBYTE(wq$$); */ + asm volatile("xvslli.b $xr8, $xr2, 1"); + asm volatile("xvslli.b $xr9, $xr3, 1"); + /* w2$$ &= NBYTES(0x1d); */ + asm volatile("xvandi.b $xr6, $xr6, 0x1d"); + asm volatile("xvandi.b $xr7, $xr7, 0x1d"); + /* w1$$ ^= w2$$; */ + asm volatile("xvxor.v $xr8, $xr8, $xr6"); + asm volatile("xvxor.v $xr9, $xr9, $xr7"); + /* wq$$ = w1$$ ^ wd$$; */ + asm volatile("xvxor.v $xr2, $xr8, $xr4"); + asm volatile("xvxor.v $xr3, $xr9, $xr5"); + } + /* *(unative_t *)&p[d+NSIZE*$$] = wp$$; */ + asm volatile("xvst $xr0, %0" : "=m"(p[d+NSIZE*0])); + asm volatile("xvst $xr1, %0" : "=m"(p[d+NSIZE*1])); + /* *(unative_t *)&q[d+NSIZE*$$] = wq$$; */ + asm volatile("xvst $xr2, %0" : "=m"(q[d+NSIZE*0])); + asm volatile("xvst $xr3, %0" : "=m"(q[d+NSIZE*1])); + } + + kernel_fpu_end(); +} + +static void raid6_lasx_xor_syndrome(int disks, int start, int stop, + size_t bytes, void **ptrs) +{ + u8 **dptr = (u8 **)ptrs; + u8 *p, *q; + int d, z, z0; + + z0 = stop; /* P/Q right side optimization */ + p = dptr[disks-2]; /* XOR parity */ + q = dptr[disks-1]; /* RS syndrome */ + + kernel_fpu_begin(); + + /* + * $xr0, $xr1: wp + * $xr2, $xr3: wq + * $xr4, $xr5: wd + * $xr6, $xr7: w2 + * $xr8, $xr9: w1 + */ + for (d = 0; d < bytes; d += NSIZE*2) { + /* P/Q data pages */ + /* wq$$ = wp$$ = *(unative_t *)&dptr[z0][d+$$*NSIZE]; */ + asm volatile("xvld $xr0, %0" : : "m"(dptr[z0][d+0*NSIZE])); + asm volatile("xvld $xr1, %0" : : "m"(dptr[z0][d+1*NSIZE])); + asm volatile("xvori.b $xr2, $xr0, 0"); + asm volatile("xvori.b $xr3, $xr1, 0"); + for (z = z0-1; z >= start; z--) { + /* wd$$ = *(unative_t *)&dptr[z][d+$$*NSIZE]; */ + asm volatile("xvld $xr4, %0" : : "m"(dptr[z][d+0*NSIZE])); + asm volatile("xvld $xr5, %0" : : "m"(dptr[z][d+1*NSIZE])); + /* wp$$ ^= wd$$; */ + asm volatile("xvxor.v $xr0, $xr0, $xr4"); + asm volatile("xvxor.v $xr1, $xr1, $xr5"); + /* w2$$ = MASK(wq$$); */ + asm volatile("xvslti.b $xr6, $xr2, 0"); + asm volatile("xvslti.b $xr7, $xr3, 0"); + /* w1$$ = SHLBYTE(wq$$); */ + asm volatile("xvslli.b $xr8, $xr2, 1"); + asm volatile("xvslli.b $xr9, $xr3, 1"); + /* w2$$ &= NBYTES(0x1d); */ + asm volatile("xvandi.b $xr6, $xr6, 0x1d"); + asm volatile("xvandi.b $xr7, $xr7, 0x1d"); + /* w1$$ ^= w2$$; */ + asm volatile("xvxor.v $xr8, $xr8, $xr6"); + asm volatile("xvxor.v $xr9, $xr9, $xr7"); + /* wq$$ = w1$$ ^ wd$$; */ + asm volatile("xvxor.v $xr2, $xr8, $xr4"); + asm volatile("xvxor.v $xr3, $xr9, $xr5"); + } + + /* P/Q left side optimization */ + for (z = start-1; z >= 0; z--) { + /* w2$$ = MASK(wq$$); */ + asm volatile("xvslti.b $xr6, $xr2, 0"); + asm volatile("xvslti.b $xr7, $xr3, 0"); + /* w1$$ = SHLBYTE(wq$$); */ + asm volatile("xvslli.b $xr8, $xr2, 1"); + asm volatile("xvslli.b $xr9, $xr3, 1"); + /* w2$$ &= NBYTES(0x1d); */ + asm volatile("xvandi.b $xr6, $xr6, 0x1d"); + asm volatile("xvandi.b $xr7, $xr7, 0x1d"); + /* wq$$ = w1$$ ^ w2$$; */ + asm volatile("xvxor.v $xr2, $xr8, $xr6"); + asm volatile("xvxor.v $xr3, $xr9, $xr7"); + } + /* + * *(unative_t *)&p[d+NSIZE*$$] ^= wp$$; + * *(unative_t *)&q[d+NSIZE*$$] ^= wq$$; + */ + asm volatile( + "xvld $xr10, %0\n\t" + "xvld $xr11, %1\n\t" + "xvld $xr12, %2\n\t" + "xvld $xr13, %3\n\t" + "xvxor.v $xr10, $xr10, $xr0\n\t" + "xvxor.v $xr11, $xr11, $xr1\n\t" + "xvxor.v $xr12, $xr12, $xr2\n\t" + "xvxor.v $xr13, $xr13, $xr3\n\t" + "xvst $xr10, %0\n\t" + "xvst $xr11, %1\n\t" + "xvst $xr12, %2\n\t" + "xvst $xr13, %3\n\t" + : "+m"(p[d+NSIZE*0]), "+m"(p[d+NSIZE*1]), + "+m"(q[d+NSIZE*0]), "+m"(q[d+NSIZE*1]) + ); + } + + kernel_fpu_end(); +} + +const struct raid6_calls raid6_lasx = { + raid6_lasx_gen_syndrome, + raid6_lasx_xor_syndrome, + raid6_has_lasx, + "lasx", + .priority = 0 /* see the comment near the top of the file for reason */ +}; +#undef NSIZE +#endif /* CONFIG_CPU_HAS_LASX */ diff --git a/lib/raid6/recov_loongarch_simd.c b/lib/raid6/recov_loongarch_simd.c new file mode 100644 index 000000000000..94aeac85e6f7 --- /dev/null +++ b/lib/raid6/recov_loongarch_simd.c @@ -0,0 +1,513 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * RAID6 recovery algorithms in LoongArch SIMD (LSX & LASX) + * + * Copyright (C) 2023 WANG Xuerui <git@xen0n.name> + * + * Originally based on recov_avx2.c and recov_ssse3.c: + * + * Copyright (C) 2012 Intel Corporation + * Author: Jim Kukunas <james.t.kukunas@linux.intel.com> + */ + +#include <linux/raid/pq.h> +#include "loongarch.h" + +/* + * Unlike with the syndrome calculation algorithms, there's no boot-time + * selection of recovery algorithms by benchmarking, so we have to specify + * the priorities and hope the future cores will all have decent vector + * support (i.e. no LASX slower than LSX, or even scalar code). + */ + +#ifdef CONFIG_CPU_HAS_LSX +static int raid6_has_lsx(void) +{ + return cpu_has_lsx; +} + +static void raid6_2data_recov_lsx(int disks, size_t bytes, int faila, + int failb, void **ptrs) +{ + u8 *p, *q, *dp, *dq; + const u8 *pbmul; /* P multiplier table for B data */ + const u8 *qmul; /* Q multiplier table (for both) */ + + p = (u8 *)ptrs[disks - 2]; + q = (u8 *)ptrs[disks - 1]; + + /* + * Compute syndrome with zero for the missing data pages + * Use the dead data pages as temporary storage for + * delta p and delta q + */ + dp = (u8 *)ptrs[faila]; + ptrs[faila] = (void *)raid6_empty_zero_page; + ptrs[disks - 2] = dp; + dq = (u8 *)ptrs[failb]; + ptrs[failb] = (void *)raid6_empty_zero_page; + ptrs[disks - 1] = dq; + + raid6_call.gen_syndrome(disks, bytes, ptrs); + + /* Restore pointer table */ + ptrs[faila] = dp; + ptrs[failb] = dq; + ptrs[disks - 2] = p; + ptrs[disks - 1] = q; + + /* Now, pick the proper data tables */ + pbmul = raid6_vgfmul[raid6_gfexi[failb - faila]]; + qmul = raid6_vgfmul[raid6_gfinv[raid6_gfexp[faila] ^ raid6_gfexp[failb]]]; + + kernel_fpu_begin(); + + /* + * vr20, vr21: qmul + * vr22, vr23: pbmul + */ + asm volatile("vld $vr20, %0" : : "m" (qmul[0])); + asm volatile("vld $vr21, %0" : : "m" (qmul[16])); + asm volatile("vld $vr22, %0" : : "m" (pbmul[0])); + asm volatile("vld $vr23, %0" : : "m" (pbmul[16])); + + while (bytes) { + /* vr4 - vr7: Q */ + asm volatile("vld $vr4, %0" : : "m" (q[0])); + asm volatile("vld $vr5, %0" : : "m" (q[16])); + asm volatile("vld $vr6, %0" : : "m" (q[32])); + asm volatile("vld $vr7, %0" : : "m" (q[48])); + /* vr4 - vr7: Q + Qxy */ + asm volatile("vld $vr8, %0" : : "m" (dq[0])); + asm volatile("vld $vr9, %0" : : "m" (dq[16])); + asm volatile("vld $vr10, %0" : : "m" (dq[32])); + asm volatile("vld $vr11, %0" : : "m" (dq[48])); + asm volatile("vxor.v $vr4, $vr4, $vr8"); + asm volatile("vxor.v $vr5, $vr5, $vr9"); + asm volatile("vxor.v $vr6, $vr6, $vr10"); + asm volatile("vxor.v $vr7, $vr7, $vr11"); + /* vr0 - vr3: P */ + asm volatile("vld $vr0, %0" : : "m" (p[0])); + asm volatile("vld $vr1, %0" : : "m" (p[16])); + asm volatile("vld $vr2, %0" : : "m" (p[32])); + asm volatile("vld $vr3, %0" : : "m" (p[48])); + /* vr0 - vr3: P + Pxy */ + asm volatile("vld $vr8, %0" : : "m" (dp[0])); + asm volatile("vld $vr9, %0" : : "m" (dp[16])); + asm volatile("vld $vr10, %0" : : "m" (dp[32])); + asm volatile("vld $vr11, %0" : : "m" (dp[48])); + asm volatile("vxor.v $vr0, $vr0, $vr8"); + asm volatile("vxor.v $vr1, $vr1, $vr9"); + asm volatile("vxor.v $vr2, $vr2, $vr10"); + asm volatile("vxor.v $vr3, $vr3, $vr11"); + + /* vr8 - vr11: higher 4 bits of each byte of (Q + Qxy) */ + asm volatile("vsrli.b $vr8, $vr4, 4"); + asm volatile("vsrli.b $vr9, $vr5, 4"); + asm volatile("vsrli.b $vr10, $vr6, 4"); + asm volatile("vsrli.b $vr11, $vr7, 4"); + /* vr4 - vr7: lower 4 bits of each byte of (Q + Qxy) */ + asm volatile("vandi.b $vr4, $vr4, 0x0f"); + asm volatile("vandi.b $vr5, $vr5, 0x0f"); + asm volatile("vandi.b $vr6, $vr6, 0x0f"); + asm volatile("vandi.b $vr7, $vr7, 0x0f"); + /* lookup from qmul[0] */ + asm volatile("vshuf.b $vr4, $vr20, $vr20, $vr4"); + asm volatile("vshuf.b $vr5, $vr20, $vr20, $vr5"); + asm volatile("vshuf.b $vr6, $vr20, $vr20, $vr6"); + asm volatile("vshuf.b $vr7, $vr20, $vr20, $vr7"); + /* lookup from qmul[16] */ + asm volatile("vshuf.b $vr8, $vr21, $vr21, $vr8"); + asm volatile("vshuf.b $vr9, $vr21, $vr21, $vr9"); + asm volatile("vshuf.b $vr10, $vr21, $vr21, $vr10"); + asm volatile("vshuf.b $vr11, $vr21, $vr21, $vr11"); + /* vr16 - vr19: B(Q + Qxy) */ + asm volatile("vxor.v $vr16, $vr8, $vr4"); + asm volatile("vxor.v $vr17, $vr9, $vr5"); + asm volatile("vxor.v $vr18, $vr10, $vr6"); + asm volatile("vxor.v $vr19, $vr11, $vr7"); + + /* vr4 - vr7: higher 4 bits of each byte of (P + Pxy) */ + asm volatile("vsrli.b $vr4, $vr0, 4"); + asm volatile("vsrli.b $vr5, $vr1, 4"); + asm volatile("vsrli.b $vr6, $vr2, 4"); + asm volatile("vsrli.b $vr7, $vr3, 4"); + /* vr12 - vr15: lower 4 bits of each byte of (P + Pxy) */ + asm volatile("vandi.b $vr12, $vr0, 0x0f"); + asm volatile("vandi.b $vr13, $vr1, 0x0f"); + asm volatile("vandi.b $vr14, $vr2, 0x0f"); + asm volatile("vandi.b $vr15, $vr3, 0x0f"); + /* lookup from pbmul[0] */ + asm volatile("vshuf.b $vr12, $vr22, $vr22, $vr12"); + asm volatile("vshuf.b $vr13, $vr22, $vr22, $vr13"); + asm volatile("vshuf.b $vr14, $vr22, $vr22, $vr14"); + asm volatile("vshuf.b $vr15, $vr22, $vr22, $vr15"); + /* lookup from pbmul[16] */ + asm volatile("vshuf.b $vr4, $vr23, $vr23, $vr4"); + asm volatile("vshuf.b $vr5, $vr23, $vr23, $vr5"); + asm volatile("vshuf.b $vr6, $vr23, $vr23, $vr6"); + asm volatile("vshuf.b $vr7, $vr23, $vr23, $vr7"); + /* vr4 - vr7: A(P + Pxy) */ + asm volatile("vxor.v $vr4, $vr4, $vr12"); + asm volatile("vxor.v $vr5, $vr5, $vr13"); + asm volatile("vxor.v $vr6, $vr6, $vr14"); + asm volatile("vxor.v $vr7, $vr7, $vr15"); + + /* vr4 - vr7: A(P + Pxy) + B(Q + Qxy) = Dx */ + asm volatile("vxor.v $vr4, $vr4, $vr16"); + asm volatile("vxor.v $vr5, $vr5, $vr17"); + asm volatile("vxor.v $vr6, $vr6, $vr18"); + asm volatile("vxor.v $vr7, $vr7, $vr19"); + asm volatile("vst $vr4, %0" : "=m" (dq[0])); + asm volatile("vst $vr5, %0" : "=m" (dq[16])); + asm volatile("vst $vr6, %0" : "=m" (dq[32])); + asm volatile("vst $vr7, %0" : "=m" (dq[48])); + + /* vr0 - vr3: P + Pxy + Dx = Dy */ + asm volatile("vxor.v $vr0, $vr0, $vr4"); + asm volatile("vxor.v $vr1, $vr1, $vr5"); + asm volatile("vxor.v $vr2, $vr2, $vr6"); + asm volatile("vxor.v $vr3, $vr3, $vr7"); + asm volatile("vst $vr0, %0" : "=m" (dp[0])); + asm volatile("vst $vr1, %0" : "=m" (dp[16])); + asm volatile("vst $vr2, %0" : "=m" (dp[32])); + asm volatile("vst $vr3, %0" : "=m" (dp[48])); + + bytes -= 64; + p += 64; + q += 64; + dp += 64; + dq += 64; + } + + kernel_fpu_end(); +} + +static void raid6_datap_recov_lsx(int disks, size_t bytes, int faila, + void **ptrs) +{ + u8 *p, *q, *dq; + const u8 *qmul; /* Q multiplier table */ + + p = (u8 *)ptrs[disks - 2]; + q = (u8 *)ptrs[disks - 1]; + + /* + * Compute syndrome with zero for the missing data page + * Use the dead data page as temporary storage for delta q + */ + dq = (u8 *)ptrs[faila]; + ptrs[faila] = (void *)raid6_empty_zero_page; + ptrs[disks - 1] = dq; + + raid6_call.gen_syndrome(disks, bytes, ptrs); + + /* Restore pointer table */ + ptrs[faila] = dq; + ptrs[disks - 1] = q; + + /* Now, pick the proper data tables */ + qmul = raid6_vgfmul[raid6_gfinv[raid6_gfexp[faila]]]; + + kernel_fpu_begin(); + + /* vr22, vr23: qmul */ + asm volatile("vld $vr22, %0" : : "m" (qmul[0])); + asm volatile("vld $vr23, %0" : : "m" (qmul[16])); + + while (bytes) { + /* vr0 - vr3: P + Dx */ + asm volatile("vld $vr0, %0" : : "m" (p[0])); + asm volatile("vld $vr1, %0" : : "m" (p[16])); + asm volatile("vld $vr2, %0" : : "m" (p[32])); + asm volatile("vld $vr3, %0" : : "m" (p[48])); + /* vr4 - vr7: Qx */ + asm volatile("vld $vr4, %0" : : "m" (dq[0])); + asm volatile("vld $vr5, %0" : : "m" (dq[16])); + asm volatile("vld $vr6, %0" : : "m" (dq[32])); + asm volatile("vld $vr7, %0" : : "m" (dq[48])); + /* vr4 - vr7: Q + Qx */ + asm volatile("vld $vr8, %0" : : "m" (q[0])); + asm volatile("vld $vr9, %0" : : "m" (q[16])); + asm volatile("vld $vr10, %0" : : "m" (q[32])); + asm volatile("vld $vr11, %0" : : "m" (q[48])); + asm volatile("vxor.v $vr4, $vr4, $vr8"); + asm volatile("vxor.v $vr5, $vr5, $vr9"); + asm volatile("vxor.v $vr6, $vr6, $vr10"); + asm volatile("vxor.v $vr7, $vr7, $vr11"); + + /* vr8 - vr11: higher 4 bits of each byte of (Q + Qx) */ + asm volatile("vsrli.b $vr8, $vr4, 4"); + asm volatile("vsrli.b $vr9, $vr5, 4"); + asm volatile("vsrli.b $vr10, $vr6, 4"); + asm volatile("vsrli.b $vr11, $vr7, 4"); + /* vr4 - vr7: lower 4 bits of each byte of (Q + Qx) */ + asm volatile("vandi.b $vr4, $vr4, 0x0f"); + asm volatile("vandi.b $vr5, $vr5, 0x0f"); + asm volatile("vandi.b $vr6, $vr6, 0x0f"); + asm volatile("vandi.b $vr7, $vr7, 0x0f"); + /* lookup from qmul[0] */ + asm volatile("vshuf.b $vr4, $vr22, $vr22, $vr4"); + asm volatile("vshuf.b $vr5, $vr22, $vr22, $vr5"); + asm volatile("vshuf.b $vr6, $vr22, $vr22, $vr6"); + asm volatile("vshuf.b $vr7, $vr22, $vr22, $vr7"); + /* lookup from qmul[16] */ + asm volatile("vshuf.b $vr8, $vr23, $vr23, $vr8"); + asm volatile("vshuf.b $vr9, $vr23, $vr23, $vr9"); + asm volatile("vshuf.b $vr10, $vr23, $vr23, $vr10"); + asm volatile("vshuf.b $vr11, $vr23, $vr23, $vr11"); + /* vr4 - vr7: qmul(Q + Qx) = Dx */ + asm volatile("vxor.v $vr4, $vr4, $vr8"); + asm volatile("vxor.v $vr5, $vr5, $vr9"); + asm volatile("vxor.v $vr6, $vr6, $vr10"); + asm volatile("vxor.v $vr7, $vr7, $vr11"); + asm volatile("vst $vr4, %0" : "=m" (dq[0])); + asm volatile("vst $vr5, %0" : "=m" (dq[16])); + asm volatile("vst $vr6, %0" : "=m" (dq[32])); + asm volatile("vst $vr7, %0" : "=m" (dq[48])); + + /* vr0 - vr3: P + Dx + Dx = P */ + asm volatile("vxor.v $vr0, $vr0, $vr4"); + asm volatile("vxor.v $vr1, $vr1, $vr5"); + asm volatile("vxor.v $vr2, $vr2, $vr6"); + asm volatile("vxor.v $vr3, $vr3, $vr7"); + asm volatile("vst $vr0, %0" : "=m" (p[0])); + asm volatile("vst $vr1, %0" : "=m" (p[16])); + asm volatile("vst $vr2, %0" : "=m" (p[32])); + asm volatile("vst $vr3, %0" : "=m" (p[48])); + + bytes -= 64; + p += 64; + q += 64; + dq += 64; + } + + kernel_fpu_end(); +} + +const struct raid6_recov_calls raid6_recov_lsx = { + .data2 = raid6_2data_recov_lsx, + .datap = raid6_datap_recov_lsx, + .valid = raid6_has_lsx, + .name = "lsx", + .priority = 1, +}; +#endif /* CONFIG_CPU_HAS_LSX */ + +#ifdef CONFIG_CPU_HAS_LASX +static int raid6_has_lasx(void) +{ + return cpu_has_lasx; +} + +static void raid6_2data_recov_lasx(int disks, size_t bytes, int faila, + int failb, void **ptrs) +{ + u8 *p, *q, *dp, *dq; + const u8 *pbmul; /* P multiplier table for B data */ + const u8 *qmul; /* Q multiplier table (for both) */ + + p = (u8 *)ptrs[disks - 2]; + q = (u8 *)ptrs[disks - 1]; + + /* + * Compute syndrome with zero for the missing data pages + * Use the dead data pages as temporary storage for + * delta p and delta q + */ + dp = (u8 *)ptrs[faila]; + ptrs[faila] = (void *)raid6_empty_zero_page; + ptrs[disks - 2] = dp; + dq = (u8 *)ptrs[failb]; + ptrs[failb] = (void *)raid6_empty_zero_page; + ptrs[disks - 1] = dq; + + raid6_call.gen_syndrome(disks, bytes, ptrs); + + /* Restore pointer table */ + ptrs[faila] = dp; + ptrs[failb] = dq; + ptrs[disks - 2] = p; + ptrs[disks - 1] = q; + + /* Now, pick the proper data tables */ + pbmul = raid6_vgfmul[raid6_gfexi[failb - faila]]; + qmul = raid6_vgfmul[raid6_gfinv[raid6_gfexp[faila] ^ raid6_gfexp[failb]]]; + + kernel_fpu_begin(); + + /* + * xr20, xr21: qmul + * xr22, xr23: pbmul + */ + asm volatile("vld $vr20, %0" : : "m" (qmul[0])); + asm volatile("vld $vr21, %0" : : "m" (qmul[16])); + asm volatile("vld $vr22, %0" : : "m" (pbmul[0])); + asm volatile("vld $vr23, %0" : : "m" (pbmul[16])); + asm volatile("xvreplve0.q $xr20, $xr20"); + asm volatile("xvreplve0.q $xr21, $xr21"); + asm volatile("xvreplve0.q $xr22, $xr22"); + asm volatile("xvreplve0.q $xr23, $xr23"); + + while (bytes) { + /* xr0, xr1: Q */ + asm volatile("xvld $xr0, %0" : : "m" (q[0])); + asm volatile("xvld $xr1, %0" : : "m" (q[32])); + /* xr0, xr1: Q + Qxy */ + asm volatile("xvld $xr4, %0" : : "m" (dq[0])); + asm volatile("xvld $xr5, %0" : : "m" (dq[32])); + asm volatile("xvxor.v $xr0, $xr0, $xr4"); + asm volatile("xvxor.v $xr1, $xr1, $xr5"); + /* xr2, xr3: P */ + asm volatile("xvld $xr2, %0" : : "m" (p[0])); + asm volatile("xvld $xr3, %0" : : "m" (p[32])); + /* xr2, xr3: P + Pxy */ + asm volatile("xvld $xr4, %0" : : "m" (dp[0])); + asm volatile("xvld $xr5, %0" : : "m" (dp[32])); + asm volatile("xvxor.v $xr2, $xr2, $xr4"); + asm volatile("xvxor.v $xr3, $xr3, $xr5"); + + /* xr4, xr5: higher 4 bits of each byte of (Q + Qxy) */ + asm volatile("xvsrli.b $xr4, $xr0, 4"); + asm volatile("xvsrli.b $xr5, $xr1, 4"); + /* xr0, xr1: lower 4 bits of each byte of (Q + Qxy) */ + asm volatile("xvandi.b $xr0, $xr0, 0x0f"); + asm volatile("xvandi.b $xr1, $xr1, 0x0f"); + /* lookup from qmul[0] */ + asm volatile("xvshuf.b $xr0, $xr20, $xr20, $xr0"); + asm volatile("xvshuf.b $xr1, $xr20, $xr20, $xr1"); + /* lookup from qmul[16] */ + asm volatile("xvshuf.b $xr4, $xr21, $xr21, $xr4"); + asm volatile("xvshuf.b $xr5, $xr21, $xr21, $xr5"); + /* xr6, xr7: B(Q + Qxy) */ + asm volatile("xvxor.v $xr6, $xr4, $xr0"); + asm volatile("xvxor.v $xr7, $xr5, $xr1"); + + /* xr4, xr5: higher 4 bits of each byte of (P + Pxy) */ + asm volatile("xvsrli.b $xr4, $xr2, 4"); + asm volatile("xvsrli.b $xr5, $xr3, 4"); + /* xr0, xr1: lower 4 bits of each byte of (P + Pxy) */ + asm volatile("xvandi.b $xr0, $xr2, 0x0f"); + asm volatile("xvandi.b $xr1, $xr3, 0x0f"); + /* lookup from pbmul[0] */ + asm volatile("xvshuf.b $xr0, $xr22, $xr22, $xr0"); + asm volatile("xvshuf.b $xr1, $xr22, $xr22, $xr1"); + /* lookup from pbmul[16] */ + asm volatile("xvshuf.b $xr4, $xr23, $xr23, $xr4"); + asm volatile("xvshuf.b $xr5, $xr23, $xr23, $xr5"); + /* xr0, xr1: A(P + Pxy) */ + asm volatile("xvxor.v $xr0, $xr0, $xr4"); + asm volatile("xvxor.v $xr1, $xr1, $xr5"); + + /* xr0, xr1: A(P + Pxy) + B(Q + Qxy) = Dx */ + asm volatile("xvxor.v $xr0, $xr0, $xr6"); + asm volatile("xvxor.v $xr1, $xr1, $xr7"); + + /* xr2, xr3: P + Pxy + Dx = Dy */ + asm volatile("xvxor.v $xr2, $xr2, $xr0"); + asm volatile("xvxor.v $xr3, $xr3, $xr1"); + + asm volatile("xvst $xr0, %0" : "=m" (dq[0])); + asm volatile("xvst $xr1, %0" : "=m" (dq[32])); + asm volatile("xvst $xr2, %0" : "=m" (dp[0])); + asm volatile("xvst $xr3, %0" : "=m" (dp[32])); + + bytes -= 64; + p += 64; + q += 64; + dp += 64; + dq += 64; + } + + kernel_fpu_end(); +} + +static void raid6_datap_recov_lasx(int disks, size_t bytes, int faila, + void **ptrs) +{ + u8 *p, *q, *dq; + const u8 *qmul; /* Q multiplier table */ + + p = (u8 *)ptrs[disks - 2]; + q = (u8 *)ptrs[disks - 1]; + + /* + * Compute syndrome with zero for the missing data page + * Use the dead data page as temporary storage for delta q + */ + dq = (u8 *)ptrs[faila]; + ptrs[faila] = (void *)raid6_empty_zero_page; + ptrs[disks - 1] = dq; + + raid6_call.gen_syndrome(disks, bytes, ptrs); + + /* Restore pointer table */ + ptrs[faila] = dq; + ptrs[disks - 1] = q; + + /* Now, pick the proper data tables */ + qmul = raid6_vgfmul[raid6_gfinv[raid6_gfexp[faila]]]; + + kernel_fpu_begin(); + + /* xr22, xr23: qmul */ + asm volatile("vld $vr22, %0" : : "m" (qmul[0])); + asm volatile("xvreplve0.q $xr22, $xr22"); + asm volatile("vld $vr23, %0" : : "m" (qmul[16])); + asm volatile("xvreplve0.q $xr23, $xr23"); + + while (bytes) { + /* xr0, xr1: P + Dx */ + asm volatile("xvld $xr0, %0" : : "m" (p[0])); + asm volatile("xvld $xr1, %0" : : "m" (p[32])); + /* xr2, xr3: Qx */ + asm volatile("xvld $xr2, %0" : : "m" (dq[0])); + asm volatile("xvld $xr3, %0" : : "m" (dq[32])); + /* xr2, xr3: Q + Qx */ + asm volatile("xvld $xr4, %0" : : "m" (q[0])); + asm volatile("xvld $xr5, %0" : : "m" (q[32])); + asm volatile("xvxor.v $xr2, $xr2, $xr4"); + asm volatile("xvxor.v $xr3, $xr3, $xr5"); + + /* xr4, xr5: higher 4 bits of each byte of (Q + Qx) */ + asm volatile("xvsrli.b $xr4, $xr2, 4"); + asm volatile("xvsrli.b $xr5, $xr3, 4"); + /* xr2, xr3: lower 4 bits of each byte of (Q + Qx) */ + asm volatile("xvandi.b $xr2, $xr2, 0x0f"); + asm volatile("xvandi.b $xr3, $xr3, 0x0f"); + /* lookup from qmul[0] */ + asm volatile("xvshuf.b $xr2, $xr22, $xr22, $xr2"); + asm volatile("xvshuf.b $xr3, $xr22, $xr22, $xr3"); + /* lookup from qmul[16] */ + asm volatile("xvshuf.b $xr4, $xr23, $xr23, $xr4"); + asm volatile("xvshuf.b $xr5, $xr23, $xr23, $xr5"); + /* xr2, xr3: qmul(Q + Qx) = Dx */ + asm volatile("xvxor.v $xr2, $xr2, $xr4"); + asm volatile("xvxor.v $xr3, $xr3, $xr5"); + + /* xr0, xr1: P + Dx + Dx = P */ + asm volatile("xvxor.v $xr0, $xr0, $xr2"); + asm volatile("xvxor.v $xr1, $xr1, $xr3"); + + asm volatile("xvst $xr2, %0" : "=m" (dq[0])); + asm volatile("xvst $xr3, %0" : "=m" (dq[32])); + asm volatile("xvst $xr0, %0" : "=m" (p[0])); + asm volatile("xvst $xr1, %0" : "=m" (p[32])); + + bytes -= 64; + p += 64; + q += 64; + dq += 64; + } + + kernel_fpu_end(); +} + +const struct raid6_recov_calls raid6_recov_lasx = { + .data2 = raid6_2data_recov_lasx, + .datap = raid6_datap_recov_lasx, + .valid = raid6_has_lasx, + .name = "lasx", + .priority = 2, +}; +#endif /* CONFIG_CPU_HAS_LASX */ diff --git a/lib/raid6/test/Makefile b/lib/raid6/test/Makefile index 1f693ea3b980..2abe0076a636 100644 --- a/lib/raid6/test/Makefile +++ b/lib/raid6/test/Makefile @@ -41,6 +41,16 @@ ifeq ($(findstring ppc,$(ARCH)),ppc) gcc -c -x c - >/dev/null && rm ./-.o && echo yes) endif +ifeq ($(ARCH),loongarch64) + CFLAGS += -I../../../arch/loongarch/include -DCONFIG_LOONGARCH=1 + CFLAGS += $(shell echo 'vld $$vr0, $$zero, 0' | \ + gcc -c -x assembler - >/dev/null 2>&1 && \ + rm ./-.o && echo -DCONFIG_CPU_HAS_LSX=1) + CFLAGS += $(shell echo 'xvld $$xr0, $$zero, 0' | \ + gcc -c -x assembler - >/dev/null 2>&1 && \ + rm ./-.o && echo -DCONFIG_CPU_HAS_LASX=1) +endif + ifeq ($(IS_X86),yes) OBJS += mmx.o sse1.o sse2.o avx2.o recov_ssse3.o recov_avx2.o avx512.o recov_avx512.o CFLAGS += -DCONFIG_X86 @@ -54,6 +64,8 @@ else ifeq ($(HAS_ALTIVEC),yes) CFLAGS += -DCONFIG_ALTIVEC OBJS += altivec1.o altivec2.o altivec4.o altivec8.o \ vpermxor1.o vpermxor2.o vpermxor4.o vpermxor8.o +else ifeq ($(ARCH),loongarch64) + OBJS += loongarch_simd.o recov_loongarch_simd.o endif .c.o: diff --git a/lib/xarray.c b/lib/xarray.c index 2071a3718f4e..39f07bfc4dcc 100644 --- a/lib/xarray.c +++ b/lib/xarray.c @@ -206,7 +206,7 @@ static void *xas_descend(struct xa_state *xas, struct xa_node *node) void *entry = xa_entry(xas->xa, node, offset); xas->xa_node = node; - if (xa_is_sibling(entry)) { + while (xa_is_sibling(entry)) { offset = xa_to_sibling(entry); entry = xa_entry(xas->xa, node, offset); if (node->shift && xa_is_node(entry)) @@ -1802,6 +1802,9 @@ EXPORT_SYMBOL(xa_get_order); * stores the index into the @id pointer, then stores the entry at * that index. A concurrent lookup will not see an uninitialised @id. * + * Must only be operated on an xarray initialized with flag XA_FLAGS_ALLOC set + * in xa_init_flags(). + * * Context: Any context. Expects xa_lock to be held on entry. May * release and reacquire xa_lock if @gfp flags permit. * Return: 0 on success, -ENOMEM if memory could not be allocated or @@ -1850,6 +1853,9 @@ EXPORT_SYMBOL(__xa_alloc); * The search for an empty entry will start at @next and will wrap * around if necessary. * + * Must only be operated on an xarray initialized with flag XA_FLAGS_ALLOC set + * in xa_init_flags(). + * * Context: Any context. Expects xa_lock to be held on entry. May * release and reacquire xa_lock if @gfp flags permit. * Return: 0 if the allocation succeeded without wrapping. 1 if the |