| Message ID | 20220802021256.162269-1-wangkefeng.wang@huawei.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | mm: memory-failure: cleanup try_to_split_thp_page() | expand |
On 2022/8/2 10:12, Kefeng Wang wrote: > Since commit 5d1fd5dc877b ("mm,hwpoison: introduce MF_MSG_UNSPLIT_THP"), > the action_result() called to show memory error event in memory_failure(), action_result(,MF_MSG_UNSPLIT_THP,) > so the pr_info() in try_to_split_thp_page() is only needed in > soft_offline_in_use_page(). > > Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com> > --- > mm/memory-failure.c | 23 ++++++++++++----------- > 1 file changed, 12 insertions(+), 11 deletions(-) > > diff --git a/mm/memory-failure.c b/mm/memory-failure.c > index f0e1961d4482..59633a617a0a 100644 > --- a/mm/memory-failure.c > +++ b/mm/memory-failure.c > @@ -1524,20 +1524,18 @@ static int identify_page_state(unsigned long pfn, struct page *p, > return page_action(ps, p, pfn); > } > > -static int try_to_split_thp_page(struct page *page, const char *msg) > +static int try_to_split_thp_page(struct page *page) > { > + int ret; > + > lock_page(page); > - if (unlikely(split_huge_page(page))) { > - unsigned long pfn = page_to_pfn(page); > + ret = split_huge_page(page); > + unlock_page(page); > > - unlock_page(page); > - pr_info("%s: %#lx: thp split failed\n", msg, pfn); > + if (unlikely(ret)) > put_page(page); > - return -EBUSY; > - } > - unlock_page(page); > > - return 0; > + return ret; > } > > static void unmap_and_kill(struct list_head *to_kill, unsigned long pfn, > @@ -2079,7 +2077,7 @@ int memory_failure(unsigned long pfn, int flags) > * page is a valid handlable page. > */ > SetPageHasHWPoisoned(hpage); > - if (try_to_split_thp_page(p, "Memory Failure") < 0) { > + if (try_to_split_thp_page(p) < 0) { > action_result(pfn, MF_MSG_UNSPLIT_THP, MF_IGNORED); > res = -EBUSY; > goto unlock_mutex; > @@ -2503,8 +2501,11 @@ static int soft_offline_in_use_page(struct page *page) > struct page *hpage = compound_head(page); > > if (!PageHuge(page) && PageTransHuge(hpage)) > - if (try_to_split_thp_page(page, "soft offline") < 0) > + if (try_to_split_thp_page(page) < 0) { > + pr_info("soft offline: %#lx: thp split failed\n", > + page_to_pfn(page)); > return -EBUSY; > + } > return __soft_offline_page(page); > } >
On Tue, Aug 02, 2022 at 10:12:56AM +0800, Kefeng Wang wrote: > Since commit 5d1fd5dc877b ("mm,hwpoison: introduce MF_MSG_UNSPLIT_THP"), > the action_result() called to show memory error event in memory_failure(), > so the pr_info() in try_to_split_thp_page() is only needed in > soft_offline_in_use_page(). > > Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com> Thanks, looks good to me. Probably this patch may come before "mm: memory-failure: convert to pr_fmt()" to be a cleanup. And recently another patch [1] is trying to change the same function, so we might need to resolve the conflict with it. I expect it's not so hard, but I think your series had better come after [1] because [1] is a bug fix and might be backported to stable-5.19. [1] https://lore.kernel.org/linux-mm/20220803025243.155798-1-fengwei.yin@intel.com/T/#u Thanks, Naoya Horiguchi > --- > mm/memory-failure.c | 23 ++++++++++++----------- > 1 file changed, 12 insertions(+), 11 deletions(-) > > diff --git a/mm/memory-failure.c b/mm/memory-failure.c > index f0e1961d4482..59633a617a0a 100644 > --- a/mm/memory-failure.c > +++ b/mm/memory-failure.c > @@ -1524,20 +1524,18 @@ static int identify_page_state(unsigned long pfn, struct page *p, > return page_action(ps, p, pfn); > } > > -static int try_to_split_thp_page(struct page *page, const char *msg) > +static int try_to_split_thp_page(struct page *page) > { > + int ret; > + > lock_page(page); > - if (unlikely(split_huge_page(page))) { > - unsigned long pfn = page_to_pfn(page); > + ret = split_huge_page(page); > + unlock_page(page); > > - unlock_page(page); > - pr_info("%s: %#lx: thp split failed\n", msg, pfn); > + if (unlikely(ret)) > put_page(page); > - return -EBUSY; > - } > - unlock_page(page); > > - return 0; > + return ret; > } > > static void unmap_and_kill(struct list_head *to_kill, unsigned long pfn, > @@ -2079,7 +2077,7 @@ int memory_failure(unsigned long pfn, int flags) > * page is a valid handlable page. > */ > SetPageHasHWPoisoned(hpage); > - if (try_to_split_thp_page(p, "Memory Failure") < 0) { > + if (try_to_split_thp_page(p) < 0) { > action_result(pfn, MF_MSG_UNSPLIT_THP, MF_IGNORED); > res = -EBUSY; > goto unlock_mutex; > @@ -2503,8 +2501,11 @@ static int soft_offline_in_use_page(struct page *page) > struct page *hpage = compound_head(page); > > if (!PageHuge(page) && PageTransHuge(hpage)) > - if (try_to_split_thp_page(page, "soft offline") < 0) > + if (try_to_split_thp_page(page) < 0) { > + pr_info("soft offline: %#lx: thp split failed\n", > + page_to_pfn(page)); > return -EBUSY; > + } > return __soft_offline_page(page); > } > > -- > 2.35.3
On 2022/8/4 8:45, HORIGUCHI NAOYA(堀口 直也) wrote: > On Tue, Aug 02, 2022 at 10:12:56AM +0800, Kefeng Wang wrote: >> Since commit 5d1fd5dc877b ("mm,hwpoison: introduce MF_MSG_UNSPLIT_THP"), >> the action_result() called to show memory error event in memory_failure(), >> so the pr_info() in try_to_split_thp_page() is only needed in >> soft_offline_in_use_page(). >> >> Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com> > Thanks, looks good to me. Probably this patch may come before > "mm: memory-failure: convert to pr_fmt()" to be a cleanup. > > And recently another patch [1] is trying to change the same function, so we > might need to resolve the conflict with it. I expect it's not so hard, but > I think your series had better come after [1] because [1] is a bug fix and > might be backported to stable-5.19. OK， I could repost after[1] merged into mm tree, thanks. > > [1] https://lore.kernel.org/linux-mm/20220803025243.155798-1-fengwei.yin@intel.com/T/#u > > Thanks, > Naoya Horiguchi
On 2022/8/4 09:03, Kefeng Wang wrote: > > On 2022/8/4 8:45, HORIGUCHI NAOYA(堀口 直也) wrote: >> On Tue, Aug 02, 2022 at 10:12:56AM +0800, Kefeng Wang wrote: >>> Since commit 5d1fd5dc877b ("mm,hwpoison: introduce MF_MSG_UNSPLIT_THP"), >>> the action_result() called to show memory error event in >>> memory_failure(), >>> so the pr_info() in try_to_split_thp_page() is only needed in >>> soft_offline_in_use_page(). >>> >>> Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com> >> Thanks, looks good to me. Probably this patch may come before >> "mm: memory-failure: convert to pr_fmt()" to be a cleanup. >> >> And recently another patch [1] is trying to change the same function, >> so we >> might need to resolve the conflict with it. I expect it's not so >> hard, but >> I think your series had better come after [1] because [1] is a bug fix >> and >> might be backported to stable-5.19. > OK， I could repost after[1] merged into mm tree, thanks. Thanks a lot for the considering merge conflict. I suppose the private data releasing will be moved to split_huge_page. There will no code conflict and not block this patch merging. Regards Yin, Fengwei >> >> [1] >> https://lore.kernel.org/linux-mm/20220803025243.155798-1-fengwei.yin@intel.com/T/#u >> >> >> Thanks, >> Naoya Horiguchi >
diff --git a/mm/memory-failure.c b/mm/memory-failure.c index f0e1961d4482..59633a617a0a 100644 --- a/mm/memory-failure.c +++ b/mm/memory-failure.c @@ -1524,20 +1524,18 @@ static int identify_page_state(unsigned long pfn, struct page *p, return page_action(ps, p, pfn); } -static int try_to_split_thp_page(struct page *page, const char *msg) +static int try_to_split_thp_page(struct page *page) { + int ret; + lock_page(page); - if (unlikely(split_huge_page(page))) { - unsigned long pfn = page_to_pfn(page); + ret = split_huge_page(page); + unlock_page(page); - unlock_page(page); - pr_info("%s: %#lx: thp split failed\n", msg, pfn); + if (unlikely(ret)) put_page(page); - return -EBUSY; - } - unlock_page(page); - return 0; + return ret; } static void unmap_and_kill(struct list_head *to_kill, unsigned long pfn, @@ -2079,7 +2077,7 @@ int memory_failure(unsigned long pfn, int flags) * page is a valid handlable page. */ SetPageHasHWPoisoned(hpage); - if (try_to_split_thp_page(p, "Memory Failure") < 0) { + if (try_to_split_thp_page(p) < 0) { action_result(pfn, MF_MSG_UNSPLIT_THP, MF_IGNORED); res = -EBUSY; goto unlock_mutex; @@ -2503,8 +2501,11 @@ static int soft_offline_in_use_page(struct page *page) struct page *hpage = compound_head(page); if (!PageHuge(page) && PageTransHuge(hpage)) - if (try_to_split_thp_page(page, "soft offline") < 0) + if (try_to_split_thp_page(page) < 0) { + pr_info("soft offline: %#lx: thp split failed\n", + page_to_pfn(page)); return -EBUSY; + } return __soft_offline_page(page); }
Since commit 5d1fd5dc877b ("mm,hwpoison: introduce MF_MSG_UNSPLIT_THP"), the action_result() called to show memory error event in memory_failure(), so the pr_info() in try_to_split_thp_page() is only needed in soft_offline_in_use_page(). Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com> --- mm/memory-failure.c | 23 ++++++++++++----------- 1 file changed, 12 insertions(+), 11 deletions(-)