Hi Baolin,
>
>
> On 8/26/26 8:24 PM, Yeoreum Yun wrote:
> > Since glibc commit 321e1fc73f (“malloc: Enable 2MB THP by default on
> > AArch64”),
> > glibc may call madvise(MADV_HUGEPAGE) for sufficiently large allocations
> > made by memalign().
> >
> > The underlying VMA may start at a different address from the aligned
> > address returned by memalign(). Furthermore, a subsequent
> > madvise(MADV_HUGEPAGE) call does not split the VMA because the flag is
> > already set.
> >
> > This causes split_huge_page_test to fail because the check_huge_xxx()
> > helpers incorrectly require the address returned by memalign() to
> > match the VMA start address reported in /proc/self/smaps.
> >
> > Fix this by using /proc/self/pagemap and /proc/kpageflags instead of
> > /proc/self/smaps to detect huge pages.
> >
> > Reported-by: David Hildenbrand (Arm) <[email protected]>
> > Signed-off-by: Yeoreum Yun <[email protected]>
> > ---
> > tools/testing/selftests/mm/vm_util.c | 130
> > ++++++++++++++++++++---------------
> > tools/testing/selftests/mm/vm_util.h | 1 +
> > 2 files changed, 77 insertions(+), 54 deletions(-)
> >
> > diff --git a/tools/testing/selftests/mm/vm_util.c
> > b/tools/testing/selftests/mm/vm_util.c
> > index 4821a3563036..1d0959b3b9e8 100644
> > --- a/tools/testing/selftests/mm/vm_util.c
> > +++ b/tools/testing/selftests/mm/vm_util.c
> > @@ -351,31 +351,13 @@ char *__get_smap_entry(void *addr, const char
> > *pattern, char *buf, size_t len)
> > return entry;
> > }
> > -static bool __check_pmd_huge(void *addr, char *pattern, int nr_hpages,
> > - uint64_t hpage_size)
> > -{
> > - char buffer[MAX_LINE_LENGTH];
> > - uint64_t thp = -1;
> > - char *entry;
> > -
> > - entry = __get_smap_entry(addr, pattern, buffer, sizeof(buffer));
> > - if (!entry)
> > - goto err_out;
> > -
> > - if (sscanf(entry, "%9" SCNu64 " kB", &thp) != 1)
> > - ksft_exit_fail_msg("Reading smap error\n");
> > -
> > -err_out:
> > - return thp == (nr_hpages * (hpage_size >> 10));
> > -}
> > -
> > -static bool check_large_folios(void *addr, size_t len, int nr_hpages,
> > - uint64_t hpage_size)
> > +static bool check_large_folios(int pagemap_fd, int kpageflags_fd,
> > + void *addr, size_t len, int nr_hpages,
> > + uint64_t hpage_size)
> > {
> > int order = 0, pagesize = getpagesize();
> > unsigned int nr_pages = hpage_size / pagesize;
> > int orders[MAX_NR_ORDERS], status;
> > - int pagemap_fd, kpageflags_fd;
> > bool ret = false;
> > if (!nr_pages)
> > @@ -386,15 +368,6 @@ static bool check_large_folios(void *addr, size_t len,
> > int nr_hpages,
> > ksft_exit_fail_msg("invalid order\n");
> > memset(orders, 0, sizeof(int) * MAX_NR_ORDERS);
> > - pagemap_fd = open(PAGEMAP_PATH, O_RDONLY);
> > - if (pagemap_fd == -1)
> > - ksft_exit_fail_msg("read pagemap fail\n");
> > -
> > - kpageflags_fd = open(KPAGEFLAGS_PATH, O_RDONLY);
> > - if (kpageflags_fd == -1) {
> > - close(pagemap_fd);
> > - ksft_exit_fail_msg("read kpageflags fail\n");
> > - }
> > status = gather_folio_orders(addr, len, pagemap_fd,
> > kpageflags_fd, orders, MAX_NR_ORDERS);
> > @@ -405,48 +378,97 @@ static bool check_large_folios(void *addr, size_t
> > len, int nr_hpages,
> > ret = true;
> > out:
> > - close(pagemap_fd);
> > - close(kpageflags_fd);
> > return ret;
> > }
> > -bool check_huge_anon(void *addr, size_t len, int nr_hpages, uint64_t
> > hpage_size)
> > +enum check_huge_type {
> > + CHECK_HUGE_ANON,
> > + CHECK_HUGE_FILE,
> > + CHECK_HUGE_SHMEM,
> > +};
> > +
> > +static bool __check_pmd_huge(void *addr, size_t len, int nr_hpages,
> > + uint64_t hpage_size, enum check_huge_type type)
>
> The original __check_pmd_huge() is only for PMD-sized large folios, but now
> it not only checks PMD-sized large folios but also mTHP large folios, which
> I find confusing. Please keep its original semantics, and only check
> PMD-sized large folios.
But, It seems to valuable to check other page-flags than checking
the large-folio only.
>
> > {
> > - uint64_t pmd_pagesize = read_pmd_pagesize();
> > + int pagemap_fd, kpageflags_fd;
> > + uint64_t pmd_pagesize, granule;
> > + uint64_t categories, kpf;
> > + unsigned long pfn;
> > + bool check_large, huge_mapped;
> > + char *start = addr;
> > + char *end = start + len;
> > + pmd_pagesize = read_pmd_pagesize();
> > if (!pmd_pagesize)
> > ksft_exit_fail_msg("reading PMD pagesize failed\n");
> > - if (hpage_size == pmd_pagesize)
> > - return __check_pmd_huge(addr, "AnonHugePages: ", nr_hpages,
> > hpage_size);
> > + if (nr_hpages > 0) {
> > + check_large = true;
> > + granule = hpage_size;
> > + } else {
> > + check_large = false;
> > + granule = psize();
> > + }
>
> This is incorrect for the mTHP large folio check. I already hit a selftest
> failure. Please test your patches before sending them out.
Since for a split case, large folio can be as-is but only remove
the PMD mapping only, skipping the large_folio checking seems valid
when nr_hpage is 0.
And the failure of test seems because of unmapped area after
changinng the collapse-order. Therefore, it seems to fine with below
patch:
---------------&<----------------------
diff --git a/tools/testing/selftests/mm/vm_util.c
b/tools/testing/selftests/mm/vm_util.c
index 1d0959b3b9e8..f174a76d2310 100644
--- a/tools/testing/selftests/mm/vm_util.c
+++ b/tools/testing/selftests/mm/vm_util.c
@@ -387,14 +387,14 @@ enum check_huge_type {
CHECK_HUGE_SHMEM,
};
-static bool __check_pmd_huge(void *addr, size_t len, int nr_hpages,
- uint64_t hpage_size, enum check_huge_type type)
+static bool __check_huge(void *addr, size_t len, int nr_hpages,
+ uint64_t hpage_size, enum check_huge_type type)
{
int pagemap_fd, kpageflags_fd;
uint64_t pmd_pagesize, granule;
uint64_t categories, kpf;
unsigned long pfn;
- bool check_large, huge_mapped;
+ bool check_large, check_huge_mapped, allow_nomap;
char *start = addr;
char *end = start + len;
@@ -405,11 +405,20 @@ static bool __check_pmd_huge(void *addr, size_t len, int
nr_hpages,
if (nr_hpages > 0) {
check_large = true;
granule = hpage_size;
+ if (granule == pmd_pagesize)
+ check_huge_mapped = true;
+ else
+ check_huge_mapped = false;
} else {
check_large = false;
granule = psize();
}
+ if (nr_hpages * hpage_size < len)
+ allow_nomap = true;
+ else
+ allow_nomap = false;
+
pagemap_fd = open(PAGEMAP_PATH, O_RDONLY);
if (pagemap_fd < 0)
ksft_exit_fail_msg("open pagemap fail\n");
@@ -428,18 +437,15 @@ static bool __check_pmd_huge(void *addr, size_t len, int
nr_hpages,
categories = pagemap_scan_get_categories(pagemap_fd, start);
pfn = pagemap_get_pfn(pagemap_fd, start);
if (pfn == -1UL) {
- if (check_large)
+ if (!allow_nomap)
goto out;
else
continue;
}
if (pageflags_get(pfn, kpageflags_fd, &kpf))
ksft_exit_fail_msg("read kpageflags: %s\n",
strerror(errno));
- huge_mapped = categories & PAGE_IS_HUGE;
- if (check_large != huge_mapped) {
- if (!check_large || granule == pmd_pagesize)
+ if (check_huge_mapped != !!(categories & PAGE_IS_HUGE))
goto out;
- }
if (kpf & KPF_COMPOUND_TAIL)
continue;
if (!!(categories & PAGE_IS_FILE) != (type != CHECK_HUGE_ANON))
@@ -458,17 +464,17 @@ static bool __check_pmd_huge(void *addr, size_t len, int
nr_hpages,
bool check_huge_anon(void *addr, size_t len, int nr_hpages, uint64_t
hpage_size)
{
- return __check_pmd_huge(addr, len, nr_hpages, hpage_size,
CHECK_HUGE_ANON);
+ return __check_huge(addr, len, nr_hpages, hpage_size, CHECK_HUGE_ANON);
}
bool check_huge_file(void *addr, size_t len, int nr_hpages, uint64_t
hpage_size)
{
- return __check_pmd_huge(addr, len, nr_hpages, hpage_size,
CHECK_HUGE_FILE);
+ return __check_huge(addr, len, nr_hpages, hpage_size, CHECK_HUGE_FILE);
}
bool check_huge_shmem(void *addr, size_t len, int nr_hpages, uint64_t
hpage_size)
{
- return __check_pmd_huge(addr, len, nr_hpages, hpage_size,
CHECK_HUGE_SHMEM);
+ return __check_huge(addr, len, nr_hpages, hpage_size, CHECK_HUGE_SHMEM);
}
If it's good for you, I'll repost again.
Thanks!
--
Sincerely,
Yeoreum Yun
