| Message ID | 20191016070924.GA10178@hori.linux.bs1.fc.nec.co.jp (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | mm, soft-offline: convert parameter to pfn | expand |
On 16.10.19 09:09, Naoya Horiguchi wrote: > Hi, > > I wrote a simple cleanup for parameter of soft_offline_page(), > based on thread https://lkml.org/lkml/2019/10/11/57. > > I know that we need more cleanup on hwpoison-inject, but I think > that will be mentioned in re-write patchset Oscar is preparing now. > So let me shared only this part as a separate one now. > > Thanks, > Naoya Horiguchi > --- > From: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> > Date: Wed, 16 Oct 2019 15:49:00 +0900 > Subject: [PATCH] mm, soft-offline: convert parameter to pfn > > Currently soft_offline_page() receives struct page, and its sibling > memory_failure() receives pfn. This discrepancy looks weird and makes > precheck on pfn validity tricky. So let's align them. > > Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> > --- > drivers/base/memory.c | 2 +- > include/linux/mm.h | 2 +- > mm/madvise.c | 2 +- > mm/memory-failure.c | 8 ++++---- > 4 files changed, 7 insertions(+), 7 deletions(-) > > diff --git a/drivers/base/memory.c b/drivers/base/memory.c > index e5485c22ef77..04e469c82852 100644 > --- a/drivers/base/memory.c > +++ b/drivers/base/memory.c > @@ -540,7 +540,7 @@ static ssize_t soft_offline_page_store(struct device *dev, > pfn >>= PAGE_SHIFT; > if (!pfn_valid(pfn)) > return -ENXIO; > - ret = soft_offline_page(pfn_to_page(pfn)); > + ret = soft_offline_page(pfn); > return ret == 0 ? count : ret; > } > > diff --git a/include/linux/mm.h b/include/linux/mm.h > index 3eba26324ff1..0a452020edf5 100644 > --- a/include/linux/mm.h > +++ b/include/linux/mm.h > @@ -2783,7 +2783,7 @@ extern int sysctl_memory_failure_early_kill; > extern int sysctl_memory_failure_recovery; > extern void shake_page(struct page *p, int access); > extern atomic_long_t num_poisoned_pages __read_mostly; > -extern int soft_offline_page(struct page *page); > +extern int soft_offline_page(unsigned long pfn); > > > /* > diff --git a/mm/madvise.c b/mm/madvise.c > index fd221b610b52..df198d1e5e2e 100644 > --- a/mm/madvise.c > +++ b/mm/madvise.c > @@ -893,7 +893,7 @@ static int madvise_inject_error(int behavior, > if (behavior == MADV_SOFT_OFFLINE) { > pr_info("Soft offlining pfn %#lx at process virtual address %#lx\n", > pfn, start); > - ret = soft_offline_page(page); > + ret = soft_offline_page(pfn); > if (ret) > return ret; > } else { > diff --git a/mm/memory-failure.c b/mm/memory-failure.c > index 4f16e0a7e7cc..eb4fd5e8d5e1 100644 > --- a/mm/memory-failure.c > +++ b/mm/memory-failure.c > @@ -1514,7 +1514,7 @@ static void memory_failure_work_func(struct work_struct *work) > if (!gotten) > break; > if (entry.flags & MF_SOFT_OFFLINE) > - soft_offline_page(pfn_to_page(entry.pfn)); > + soft_offline_page(entry.pfn); > else > memory_failure(entry.pfn, entry.flags); > } > @@ -1822,7 +1822,7 @@ static int soft_offline_free_page(struct page *page) > > /** > * soft_offline_page - Soft offline a page. > - * @page: page to offline > + * @pfn: pfn to soft-offline > * > * Returns 0 on success, otherwise negated errno. > * > @@ -1841,10 +1841,10 @@ static int soft_offline_free_page(struct page *page) > * This is not a 100% solution for all memory, but tries to be > * ``good enough'' for the majority of memory. > */ > -int soft_offline_page(struct page *page) > +int soft_offline_page(unsigned long pfn) > { > int ret; > - unsigned long pfn = page_to_pfn(page); > + struct page *page = pfn_to_page(pfn); > > if (is_zone_device_page(page)) { > pr_debug_ratelimited("soft_offline: %#lx page is device page\n", > I think you should rebase that patch on linux-next (where the pfn_to_online_page() check is in place). I assume you'll want to move the pfn_to_online_page() check into soft_offline_page() then as well?
On Wed, Oct 16, 2019 at 09:56:19AM +0200, David Hildenbrand wrote: > On 16.10.19 09:09, Naoya Horiguchi wrote: > > Hi, > > > > I wrote a simple cleanup for parameter of soft_offline_page(), > > based on thread https://lkml.org/lkml/2019/10/11/57. > > > > I know that we need more cleanup on hwpoison-inject, but I think > > that will be mentioned in re-write patchset Oscar is preparing now. > > So let me shared only this part as a separate one now. ... > > I think you should rebase that patch on linux-next (where the > pfn_to_online_page() check is in place). I assume you'll want to move the > pfn_to_online_page() check into soft_offline_page() then as well? I rebased to next-20191016. And yes, we will move pfn_to_online_page() into soft offline code. It seems that we can also move pfn_valid(), but is simply moving like below good enough for you? @@ -1877,11 +1877,17 @@ static int soft_offline_free_page(struct page *page) * This is not a 100% solution for all memory, but tries to be * ``good enough'' for the majority of memory. */ -int soft_offline_page(struct page *page, int flags) +int soft_offline_page(unsigned long pfn, int flags) { int ret; - unsigned long pfn = page_to_pfn(page); + struct page *page; + if (!pfn_valid(pfn)) + return -ENXIO; + /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ + if (!pfn_to_online_page(pfn)) + return -EIO; + page = pfn_to_page(pfn); if (is_zone_device_page(page)) { pr_debug_ratelimited("soft_offline: %#lx page is device page\n", pfn); -- Or we might have an option to do as memory_failure() does like below: int memory_failure(unsigned long pfn, int flags) { .... p = pfn_to_online_page(pfn); if (!p) { if (pfn_valid(pfn)) { pgmap = get_dev_pagemap(pfn, NULL); if (pgmap) return memory_failure_dev_pagemap(pfn, flags, pgmap); } pr_err("Memory failure: %#lx: memory outside kernel control\n", pfn); return -ENXIO; } Thanks, Naoya Horiguchi
On 16.10.19 10:27, Naoya Horiguchi wrote: > On Wed, Oct 16, 2019 at 09:56:19AM +0200, David Hildenbrand wrote: >> On 16.10.19 09:09, Naoya Horiguchi wrote: >>> Hi, >>> >>> I wrote a simple cleanup for parameter of soft_offline_page(), >>> based on thread https://lkml.org/lkml/2019/10/11/57. >>> >>> I know that we need more cleanup on hwpoison-inject, but I think >>> that will be mentioned in re-write patchset Oscar is preparing now. >>> So let me shared only this part as a separate one now. > ... >> >> I think you should rebase that patch on linux-next (where the >> pfn_to_online_page() check is in place). I assume you'll want to move the >> pfn_to_online_page() check into soft_offline_page() then as well? > > I rebased to next-20191016. And yes, we will move pfn_to_online_page() > into soft offline code. It seems that we can also move pfn_valid(), > but is simply moving like below good enough for you? At least I can't am the patch to current next/master (due to pfn_to_online_page()). > > @@ -1877,11 +1877,17 @@ static int soft_offline_free_page(struct page *page) > * This is not a 100% solution for all memory, but tries to be > * ``good enough'' for the majority of memory. > */ > -int soft_offline_page(struct page *page, int flags) > +int soft_offline_page(unsigned long pfn, int flags) > { > int ret; > - unsigned long pfn = page_to_pfn(page); > + struct page *page; > > + if (!pfn_valid(pfn)) > + return -ENXIO; > + /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ > + if (!pfn_to_online_page(pfn)) > + return -EIO; > + page = pfn_to_page(pfn); > if (is_zone_device_page(page)) { > pr_debug_ratelimited("soft_offline: %#lx page is device page\n", > pfn); > -- > > Or we might have an option to do as memory_failure() does like below: In contrast to soft offlining, memory failure can deal with devmem. So I think the above makes sense.
On Wed, Oct 16, 2019 at 10:34:52AM +0200, David Hildenbrand wrote: > On 16.10.19 10:27, Naoya Horiguchi wrote: > > On Wed, Oct 16, 2019 at 09:56:19AM +0200, David Hildenbrand wrote: > > > On 16.10.19 09:09, Naoya Horiguchi wrote: > > > > Hi, > > > > > > > > I wrote a simple cleanup for parameter of soft_offline_page(), > > > > based on thread https://lkml.org/lkml/2019/10/11/57. > > > > > > > > I know that we need more cleanup on hwpoison-inject, but I think > > > > that will be mentioned in re-write patchset Oscar is preparing now. > > > > So let me shared only this part as a separate one now. > > ... > > > > > > I think you should rebase that patch on linux-next (where the > > > pfn_to_online_page() check is in place). I assume you'll want to move the > > > pfn_to_online_page() check into soft_offline_page() then as well? > > > > I rebased to next-20191016. And yes, we will move pfn_to_online_page() > > into soft offline code. It seems that we can also move pfn_valid(), > > but is simply moving like below good enough for you? > > At least I can't am the patch to current next/master (due to > pfn_to_online_page()). > > > > > @@ -1877,11 +1877,17 @@ static int soft_offline_free_page(struct page *page) > > * This is not a 100% solution for all memory, but tries to be > > * ``good enough'' for the majority of memory. > > */ > > -int soft_offline_page(struct page *page, int flags) > > +int soft_offline_page(unsigned long pfn, int flags) > > { > > int ret; > > - unsigned long pfn = page_to_pfn(page); > > + struct page *page; > > + if (!pfn_valid(pfn)) > > + return -ENXIO; > > + /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ > > + if (!pfn_to_online_page(pfn)) > > + return -EIO; > > + page = pfn_to_page(pfn); > > if (is_zone_device_page(page)) { > > pr_debug_ratelimited("soft_offline: %#lx page is device page\n", > > pfn); > > -- > > > > Or we might have an option to do as memory_failure() does like below: > > In contrast to soft offlining, memory failure can deal with devmem. So I > think the above makes sense. OK, so here's the revised one. Thanks, Naoya Horiguchi --- From: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Date: Wed, 16 Oct 2019 17:00:33 +0900 Subject: [PATCH] mm, soft-offline: convert parameter to pfn Currently soft_offline_page() receives struct page, and its sibling memory_failure() receives pfn. This discrepancy looks weird and makes precheck on pfn validity tricky. So let's align them. Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> --- drivers/base/memory.c | 7 +------ include/linux/mm.h | 2 +- mm/madvise.c | 2 +- mm/memory-failure.c | 14 ++++++++++---- 4 files changed, 13 insertions(+), 12 deletions(-) diff --git a/drivers/base/memory.c b/drivers/base/memory.c index 55907c27075b..a757d9ed88a7 100644 --- a/drivers/base/memory.c +++ b/drivers/base/memory.c @@ -538,12 +538,7 @@ static ssize_t soft_offline_page_store(struct device *dev, if (kstrtoull(buf, 0, &pfn) < 0) return -EINVAL; pfn >>= PAGE_SHIFT; - if (!pfn_valid(pfn)) - return -ENXIO; - /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ - if (!pfn_to_online_page(pfn)) - return -EIO; - ret = soft_offline_page(pfn_to_page(pfn), 0); + ret = soft_offline_page(pfn, 0); return ret == 0 ? count : ret; } diff --git a/include/linux/mm.h b/include/linux/mm.h index 44d058723db9..fd360d208346 100644 --- a/include/linux/mm.h +++ b/include/linux/mm.h @@ -2794,7 +2794,7 @@ extern int sysctl_memory_failure_early_kill; extern int sysctl_memory_failure_recovery; extern void shake_page(struct page *p, int access); extern atomic_long_t num_poisoned_pages __read_mostly; -extern int soft_offline_page(struct page *page, int flags); +extern int soft_offline_page(unsigned long pfn, int flags); /* diff --git a/mm/madvise.c b/mm/madvise.c index 2be9f3fdb05e..99dd06fecfa9 100644 --- a/mm/madvise.c +++ b/mm/madvise.c @@ -887,7 +887,7 @@ static int madvise_inject_error(int behavior, pr_info("Soft offlining pfn %#lx at process virtual address %#lx\n", pfn, start); - ret = soft_offline_page(page, MF_COUNT_INCREASED); + ret = soft_offline_page(pfn, MF_COUNT_INCREASED); if (ret) return ret; continue; diff --git a/mm/memory-failure.c b/mm/memory-failure.c index 05c8c6df25e6..bdf408d7f65c 100644 --- a/mm/memory-failure.c +++ b/mm/memory-failure.c @@ -1476,7 +1476,7 @@ static void memory_failure_work_func(struct work_struct *work) if (!gotten) break; if (entry.flags & MF_SOFT_OFFLINE) - soft_offline_page(pfn_to_page(entry.pfn), entry.flags); + soft_offline_page(entry.pfn, entry.flags); else memory_failure(entry.pfn, entry.flags); } @@ -1857,7 +1857,7 @@ static int soft_offline_free_page(struct page *page) /** * soft_offline_page - Soft offline a page. - * @page: page to offline + * @pfn: pfn to soft-offline * @flags: flags. Same as memory_failure(). * * Returns 0 on success, otherwise negated errno. @@ -1877,11 +1877,17 @@ static int soft_offline_free_page(struct page *page) * This is not a 100% solution for all memory, but tries to be * ``good enough'' for the majority of memory. */ -int soft_offline_page(struct page *page, int flags) +int soft_offline_page(unsigned long pfn, int flags) { int ret; - unsigned long pfn = page_to_pfn(page); + struct page *page; + if (!pfn_valid(pfn)) + return -ENXIO; + /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ + page = pfn_to_online_page(pfn); + if (!page) + return -EIO; if (is_zone_device_page(page)) { pr_debug_ratelimited("soft_offline: %#lx page is device page\n", pfn);
On 16.10.19 10:54, Naoya Horiguchi wrote: > On Wed, Oct 16, 2019 at 10:34:52AM +0200, David Hildenbrand wrote: >> On 16.10.19 10:27, Naoya Horiguchi wrote: >>> On Wed, Oct 16, 2019 at 09:56:19AM +0200, David Hildenbrand wrote: >>>> On 16.10.19 09:09, Naoya Horiguchi wrote: >>>>> Hi, >>>>> >>>>> I wrote a simple cleanup for parameter of soft_offline_page(), >>>>> based on thread https://lkml.org/lkml/2019/10/11/57. >>>>> >>>>> I know that we need more cleanup on hwpoison-inject, but I think >>>>> that will be mentioned in re-write patchset Oscar is preparing now. >>>>> So let me shared only this part as a separate one now. >>> ... >>>> >>>> I think you should rebase that patch on linux-next (where the >>>> pfn_to_online_page() check is in place). I assume you'll want to move the >>>> pfn_to_online_page() check into soft_offline_page() then as well? >>> >>> I rebased to next-20191016. And yes, we will move pfn_to_online_page() >>> into soft offline code. It seems that we can also move pfn_valid(), >>> but is simply moving like below good enough for you? >> >> At least I can't am the patch to current next/master (due to >> pfn_to_online_page()). Could also be that my "git am" skills failed as the mail was not a proper patch itself :) >> >>> >>> @@ -1877,11 +1877,17 @@ static int soft_offline_free_page(struct page *page) >>> * This is not a 100% solution for all memory, but tries to be >>> * ``good enough'' for the majority of memory. >>> */ >>> -int soft_offline_page(struct page *page, int flags) >>> +int soft_offline_page(unsigned long pfn, int flags) >>> { >>> int ret; >>> - unsigned long pfn = page_to_pfn(page); >>> + struct page *page; >>> + if (!pfn_valid(pfn)) >>> + return -ENXIO; >>> + /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ >>> + if (!pfn_to_online_page(pfn)) >>> + return -EIO; >>> + page = pfn_to_page(pfn); >>> if (is_zone_device_page(page)) { >>> pr_debug_ratelimited("soft_offline: %#lx page is device page\n", >>> pfn); >>> -- >>> >>> Or we might have an option to do as memory_failure() does like below: >> >> In contrast to soft offlining, memory failure can deal with devmem. So I >> think the above makes sense. > > OK, so here's the revised one. > > Thanks, > Naoya Horiguchi > --- > From: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> > Date: Wed, 16 Oct 2019 17:00:33 +0900 > Subject: [PATCH] mm, soft-offline: convert parameter to pfn > > Currently soft_offline_page() receives struct page, and its sibling > memory_failure() receives pfn. This discrepancy looks weird and makes > precheck on pfn validity tricky. So let's align them. > > Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> > --- > drivers/base/memory.c | 7 +------ > include/linux/mm.h | 2 +- > mm/madvise.c | 2 +- > mm/memory-failure.c | 14 ++++++++++---- > 4 files changed, 13 insertions(+), 12 deletions(-) > > diff --git a/drivers/base/memory.c b/drivers/base/memory.c > index 55907c27075b..a757d9ed88a7 100644 > --- a/drivers/base/memory.c > +++ b/drivers/base/memory.c > @@ -538,12 +538,7 @@ static ssize_t soft_offline_page_store(struct device *dev, > if (kstrtoull(buf, 0, &pfn) < 0) > return -EINVAL; > pfn >>= PAGE_SHIFT; > - if (!pfn_valid(pfn)) > - return -ENXIO; > - /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ > - if (!pfn_to_online_page(pfn)) > - return -EIO; > - ret = soft_offline_page(pfn_to_page(pfn), 0); > + ret = soft_offline_page(pfn, 0); > return ret == 0 ? count : ret; > } > > diff --git a/include/linux/mm.h b/include/linux/mm.h > index 44d058723db9..fd360d208346 100644 > --- a/include/linux/mm.h > +++ b/include/linux/mm.h > @@ -2794,7 +2794,7 @@ extern int sysctl_memory_failure_early_kill; > extern int sysctl_memory_failure_recovery; > extern void shake_page(struct page *p, int access); > extern atomic_long_t num_poisoned_pages __read_mostly; > -extern int soft_offline_page(struct page *page, int flags); > +extern int soft_offline_page(unsigned long pfn, int flags); > > > /* > diff --git a/mm/madvise.c b/mm/madvise.c > index 2be9f3fdb05e..99dd06fecfa9 100644 > --- a/mm/madvise.c > +++ b/mm/madvise.c > @@ -887,7 +887,7 @@ static int madvise_inject_error(int behavior, > pr_info("Soft offlining pfn %#lx at process virtual address %#lx\n", > pfn, start); > > - ret = soft_offline_page(page, MF_COUNT_INCREASED); > + ret = soft_offline_page(pfn, MF_COUNT_INCREASED); > if (ret) > return ret; > continue; > diff --git a/mm/memory-failure.c b/mm/memory-failure.c > index 05c8c6df25e6..bdf408d7f65c 100644 > --- a/mm/memory-failure.c > +++ b/mm/memory-failure.c > @@ -1476,7 +1476,7 @@ static void memory_failure_work_func(struct work_struct *work) > if (!gotten) > break; > if (entry.flags & MF_SOFT_OFFLINE) > - soft_offline_page(pfn_to_page(entry.pfn), entry.flags); > + soft_offline_page(entry.pfn, entry.flags); > else > memory_failure(entry.pfn, entry.flags); > } > @@ -1857,7 +1857,7 @@ static int soft_offline_free_page(struct page *page) > > /** > * soft_offline_page - Soft offline a page. > - * @page: page to offline > + * @pfn: pfn to soft-offline > * @flags: flags. Same as memory_failure(). > * > * Returns 0 on success, otherwise negated errno. > @@ -1877,11 +1877,17 @@ static int soft_offline_free_page(struct page *page) > * This is not a 100% solution for all memory, but tries to be > * ``good enough'' for the majority of memory. > */ > -int soft_offline_page(struct page *page, int flags) > +int soft_offline_page(unsigned long pfn, int flags) > { > int ret; > - unsigned long pfn = page_to_pfn(page); > + struct page *page; > > + if (!pfn_valid(pfn)) > + return -ENXIO; > + /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ > + page = pfn_to_online_page(pfn); > + if (!page) > + return -EIO; > if (is_zone_device_page(page)) { -> this is now no longer possible! So you can drop the whole if (is_zone_device....) case > pr_debug_ratelimited("soft_offline: %#lx page is device page\n", > pfn); > Apart from that, looks good to me.
On Wed, Oct 16, 2019 at 10:57:57AM +0200, David Hildenbrand wrote: > On 16.10.19 10:54, Naoya Horiguchi wrote: > >On Wed, Oct 16, 2019 at 10:34:52AM +0200, David Hildenbrand wrote: > >>On 16.10.19 10:27, Naoya Horiguchi wrote: > >>>On Wed, Oct 16, 2019 at 09:56:19AM +0200, David Hildenbrand wrote: > >>>>On 16.10.19 09:09, Naoya Horiguchi wrote: > >>>>>Hi, > >>>>> > >>>>>I wrote a simple cleanup for parameter of soft_offline_page(), > >>>>>based on thread https://lkml.org/lkml/2019/10/11/57. > >>>>> > >>>>>I know that we need more cleanup on hwpoison-inject, but I think > >>>>>that will be mentioned in re-write patchset Oscar is preparing now. > >>>>>So let me shared only this part as a separate one now. > >>>... > >>>> > >>>>I think you should rebase that patch on linux-next (where the > >>>>pfn_to_online_page() check is in place). I assume you'll want to move the > >>>>pfn_to_online_page() check into soft_offline_page() then as well? > >>> > >>>I rebased to next-20191016. And yes, we will move pfn_to_online_page() > >>>into soft offline code. It seems that we can also move pfn_valid(), > >>>but is simply moving like below good enough for you? > >> > >>At least I can't am the patch to current next/master (due to > >>pfn_to_online_page()). > > Could also be that my "git am" skills failed as the mail was not a > proper patch itself :) Sorry for the inconvenience, my company email system breaks original message by introducing quoted-printable format ('=20' or '=3D'). Most mail client usually handles it but git-am doesn't. I give up using it and send via smtp.gmail.com. > >@@ -1877,11 +1877,17 @@ static int soft_offline_free_page(struct page *page) > > * This is not a 100% solution for all memory, but tries to be > > * ``good enough'' for the majority of memory. > > */ > >-int soft_offline_page(struct page *page, int flags) > >+int soft_offline_page(unsigned long pfn, int flags) > > { > > int ret; > >- unsigned long pfn = page_to_pfn(page); > >+ struct page *page; > >+ if (!pfn_valid(pfn)) > >+ return -ENXIO; > >+ /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ > >+ page = pfn_to_online_page(pfn); > >+ if (!page) > >+ return -EIO; > > if (is_zone_device_page(page)) { > > -> this is now no longer possible! So you can drop the whole if > (is_zone_device....) case OK, thanks. I updated it. Thanks, Naoya Horiguchi --- From 5faf227839b578726fe7f5ff414a153abb3b3a31 Mon Sep 17 00:00:00 2001 From: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Date: Thu, 17 Oct 2019 08:40:53 +0900 Subject: [PATCH] mm, soft-offline: convert parameter to pfn Currently soft_offline_page() receives struct page, and its sibling memory_failure() receives pfn. This discrepancy looks weird and makes precheck on pfn validity tricky. So let's align them. Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> --- drivers/base/memory.c | 7 +------ include/linux/mm.h | 2 +- mm/madvise.c | 2 +- mm/memory-failure.c | 19 +++++++++---------- 4 files changed, 12 insertions(+), 18 deletions(-) diff --git a/drivers/base/memory.c b/drivers/base/memory.c index 55907c27075b..a757d9ed88a7 100644 --- a/drivers/base/memory.c +++ b/drivers/base/memory.c @@ -538,12 +538,7 @@ static ssize_t soft_offline_page_store(struct device *dev, if (kstrtoull(buf, 0, &pfn) < 0) return -EINVAL; pfn >>= PAGE_SHIFT; - if (!pfn_valid(pfn)) - return -ENXIO; - /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ - if (!pfn_to_online_page(pfn)) - return -EIO; - ret = soft_offline_page(pfn_to_page(pfn), 0); + ret = soft_offline_page(pfn, 0); return ret == 0 ? count : ret; } diff --git a/include/linux/mm.h b/include/linux/mm.h index 44d058723db9..fd360d208346 100644 --- a/include/linux/mm.h +++ b/include/linux/mm.h @@ -2794,7 +2794,7 @@ extern int sysctl_memory_failure_early_kill; extern int sysctl_memory_failure_recovery; extern void shake_page(struct page *p, int access); extern atomic_long_t num_poisoned_pages __read_mostly; -extern int soft_offline_page(struct page *page, int flags); +extern int soft_offline_page(unsigned long pfn, int flags); /* diff --git a/mm/madvise.c b/mm/madvise.c index 2be9f3fdb05e..99dd06fecfa9 100644 --- a/mm/madvise.c +++ b/mm/madvise.c @@ -887,7 +887,7 @@ static int madvise_inject_error(int behavior, pr_info("Soft offlining pfn %#lx at process virtual address %#lx\n", pfn, start); - ret = soft_offline_page(page, MF_COUNT_INCREASED); + ret = soft_offline_page(pfn, MF_COUNT_INCREASED); if (ret) return ret; continue; diff --git a/mm/memory-failure.c b/mm/memory-failure.c index 05c8c6df25e6..af2712004a4d 100644 --- a/mm/memory-failure.c +++ b/mm/memory-failure.c @@ -1476,7 +1476,7 @@ static void memory_failure_work_func(struct work_struct *work) if (!gotten) break; if (entry.flags & MF_SOFT_OFFLINE) - soft_offline_page(pfn_to_page(entry.pfn), entry.flags); + soft_offline_page(entry.pfn, entry.flags); else memory_failure(entry.pfn, entry.flags); } @@ -1857,7 +1857,7 @@ static int soft_offline_free_page(struct page *page) /** * soft_offline_page - Soft offline a page. - * @page: page to offline + * @pfn: pfn to soft-offline * @flags: flags. Same as memory_failure(). * * Returns 0 on success, otherwise negated errno. @@ -1877,18 +1877,17 @@ static int soft_offline_free_page(struct page *page) * This is not a 100% solution for all memory, but tries to be * ``good enough'' for the majority of memory. */ -int soft_offline_page(struct page *page, int flags) +int soft_offline_page(unsigned long pfn, int flags) { int ret; - unsigned long pfn = page_to_pfn(page); + struct page *page; - if (is_zone_device_page(page)) { - pr_debug_ratelimited("soft_offline: %#lx page is device page\n", - pfn); - if (flags & MF_COUNT_INCREASED) - put_page(page); + if (!pfn_valid(pfn)) + return -ENXIO; + /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ + page = pfn_to_online_page(pfn); + if (!page) return -EIO; - } if (PageHWPoison(page)) { pr_info("soft offline: %#lx page already poisoned\n", pfn);
On 17.10.19 01:47, Naoya Horiguchi wrote: > On Wed, Oct 16, 2019 at 10:57:57AM +0200, David Hildenbrand wrote: >> On 16.10.19 10:54, Naoya Horiguchi wrote: >>> On Wed, Oct 16, 2019 at 10:34:52AM +0200, David Hildenbrand wrote: >>>> On 16.10.19 10:27, Naoya Horiguchi wrote: >>>>> On Wed, Oct 16, 2019 at 09:56:19AM +0200, David Hildenbrand wrote: >>>>>> On 16.10.19 09:09, Naoya Horiguchi wrote: >>>>>>> Hi, >>>>>>> >>>>>>> I wrote a simple cleanup for parameter of soft_offline_page(), >>>>>>> based on thread https://lkml.org/lkml/2019/10/11/57. >>>>>>> >>>>>>> I know that we need more cleanup on hwpoison-inject, but I think >>>>>>> that will be mentioned in re-write patchset Oscar is preparing now. >>>>>>> So let me shared only this part as a separate one now. >>>>> ... >>>>>> >>>>>> I think you should rebase that patch on linux-next (where the >>>>>> pfn_to_online_page() check is in place). I assume you'll want to move the >>>>>> pfn_to_online_page() check into soft_offline_page() then as well? >>>>> >>>>> I rebased to next-20191016. And yes, we will move pfn_to_online_page() >>>>> into soft offline code. It seems that we can also move pfn_valid(), >>>>> but is simply moving like below good enough for you? >>>> >>>> At least I can't am the patch to current next/master (due to >>>> pfn_to_online_page()). >> >> Could also be that my "git am" skills failed as the mail was not a >> proper patch itself :) > > Sorry for the inconvenience, my company email system breaks original > message by introducing quoted-printable format ('=20' or '=3D'). > Most mail client usually handles it but git-am doesn't. > I give up using it and send via smtp.gmail.com. > >>> @@ -1877,11 +1877,17 @@ static int soft_offline_free_page(struct page *page) >>> * This is not a 100% solution for all memory, but tries to be >>> * ``good enough'' for the majority of memory. >>> */ >>> -int soft_offline_page(struct page *page, int flags) >>> +int soft_offline_page(unsigned long pfn, int flags) >>> { >>> int ret; >>> - unsigned long pfn = page_to_pfn(page); >>> + struct page *page; >>> + if (!pfn_valid(pfn)) >>> + return -ENXIO; >>> + /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ >>> + page = pfn_to_online_page(pfn); >>> + if (!page) >>> + return -EIO; >>> if (is_zone_device_page(page)) { >> >> -> this is now no longer possible! So you can drop the whole if >> (is_zone_device....) case > > OK, thanks. I updated it. > > Thanks, > Naoya Horiguchi > --- > From 5faf227839b578726fe7f5ff414a153abb3b3a31 Mon Sep 17 00:00:00 2001 > From: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> > Date: Thu, 17 Oct 2019 08:40:53 +0900 > Subject: [PATCH] mm, soft-offline: convert parameter to pfn > > Currently soft_offline_page() receives struct page, and its sibling > memory_failure() receives pfn. This discrepancy looks weird and makes > precheck on pfn validity tricky. So let's align them. > > Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> > --- > drivers/base/memory.c | 7 +------ > include/linux/mm.h | 2 +- > mm/madvise.c | 2 +- > mm/memory-failure.c | 19 +++++++++---------- > 4 files changed, 12 insertions(+), 18 deletions(-) > > diff --git a/drivers/base/memory.c b/drivers/base/memory.c > index 55907c27075b..a757d9ed88a7 100644 > --- a/drivers/base/memory.c > +++ b/drivers/base/memory.c > @@ -538,12 +538,7 @@ static ssize_t soft_offline_page_store(struct device *dev, > if (kstrtoull(buf, 0, &pfn) < 0) > return -EINVAL; > pfn >>= PAGE_SHIFT; > - if (!pfn_valid(pfn)) > - return -ENXIO; > - /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ > - if (!pfn_to_online_page(pfn)) > - return -EIO; > - ret = soft_offline_page(pfn_to_page(pfn), 0); > + ret = soft_offline_page(pfn, 0); > return ret == 0 ? count : ret; > } > > diff --git a/include/linux/mm.h b/include/linux/mm.h > index 44d058723db9..fd360d208346 100644 > --- a/include/linux/mm.h > +++ b/include/linux/mm.h > @@ -2794,7 +2794,7 @@ extern int sysctl_memory_failure_early_kill; > extern int sysctl_memory_failure_recovery; > extern void shake_page(struct page *p, int access); > extern atomic_long_t num_poisoned_pages __read_mostly; > -extern int soft_offline_page(struct page *page, int flags); > +extern int soft_offline_page(unsigned long pfn, int flags); > > > /* > diff --git a/mm/madvise.c b/mm/madvise.c > index 2be9f3fdb05e..99dd06fecfa9 100644 > --- a/mm/madvise.c > +++ b/mm/madvise.c > @@ -887,7 +887,7 @@ static int madvise_inject_error(int behavior, > pr_info("Soft offlining pfn %#lx at process virtual address %#lx\n", > pfn, start); > > - ret = soft_offline_page(page, MF_COUNT_INCREASED); > + ret = soft_offline_page(pfn, MF_COUNT_INCREASED); > if (ret) > return ret; > continue; > diff --git a/mm/memory-failure.c b/mm/memory-failure.c > index 05c8c6df25e6..af2712004a4d 100644 > --- a/mm/memory-failure.c > +++ b/mm/memory-failure.c > @@ -1476,7 +1476,7 @@ static void memory_failure_work_func(struct work_struct *work) > if (!gotten) > break; > if (entry.flags & MF_SOFT_OFFLINE) > - soft_offline_page(pfn_to_page(entry.pfn), entry.flags); > + soft_offline_page(entry.pfn, entry.flags); > else > memory_failure(entry.pfn, entry.flags); > } > @@ -1857,7 +1857,7 @@ static int soft_offline_free_page(struct page *page) > > /** > * soft_offline_page - Soft offline a page. > - * @page: page to offline > + * @pfn: pfn to soft-offline > * @flags: flags. Same as memory_failure(). > * > * Returns 0 on success, otherwise negated errno. > @@ -1877,18 +1877,17 @@ static int soft_offline_free_page(struct page *page) > * This is not a 100% solution for all memory, but tries to be > * ``good enough'' for the majority of memory. > */ > -int soft_offline_page(struct page *page, int flags) > +int soft_offline_page(unsigned long pfn, int flags) > { > int ret; > - unsigned long pfn = page_to_pfn(page); > + struct page *page; > > - if (is_zone_device_page(page)) { > - pr_debug_ratelimited("soft_offline: %#lx page is device page\n", > - pfn); > - if (flags & MF_COUNT_INCREASED) > - put_page(page); > + if (!pfn_valid(pfn)) > + return -ENXIO; > + /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ > + page = pfn_to_online_page(pfn); > + if (!page) If you pass in a PFN with MF_COUNT_INCREASED via mm/madvise.c, you would now no longer do a put_page(page) in case of ZONE_DEVICE (!page = pfn_to_online_page(pfn);) something like this page = pfn_to_online_page(pfn); if (!page) { /* * With MF_COUNT_INCREASED, we can use pfn_to_page() directly * (esp., ZONE_DEVICE). */ if (flags & MF_COUNT_INCREASED) put_page(pfn_to_page(page)); return -EIO; } For !pfn_valid(pfn), this is not relevant. > return -EIO; > - } > > if (PageHWPoison(page)) { > pr_info("soft offline: %#lx page already poisoned\n", pfn); >
On Thu, Oct 17, 2019 at 09:16:42AM +0200, David Hildenbrand wrote: > On 17.10.19 01:47, Naoya Horiguchi wrote: > > On Wed, Oct 16, 2019 at 10:57:57AM +0200, David Hildenbrand wrote: > > > On 16.10.19 10:54, Naoya Horiguchi wrote: > > > > On Wed, Oct 16, 2019 at 10:34:52AM +0200, David Hildenbrand wrote: > > > > > On 16.10.19 10:27, Naoya Horiguchi wrote: > > > > > > On Wed, Oct 16, 2019 at 09:56:19AM +0200, David Hildenbrand wrote: > > > > > > > On 16.10.19 09:09, Naoya Horiguchi wrote: > > > > > > > > Hi, > > > > > > > > > > > > > > > > I wrote a simple cleanup for parameter of soft_offline_page(), > > > > > > > > based on thread https://lkml.org/lkml/2019/10/11/57. > > > > > > > > > > > > > > > > I know that we need more cleanup on hwpoison-inject, but I think > > > > > > > > that will be mentioned in re-write patchset Oscar is preparing now. > > > > > > > > So let me shared only this part as a separate one now. > > > > > > ... > > > > > > > > > > > > > > I think you should rebase that patch on linux-next (where the > > > > > > > pfn_to_online_page() check is in place). I assume you'll want to move the > > > > > > > pfn_to_online_page() check into soft_offline_page() then as well? > > > > > > > > > > > > I rebased to next-20191016. And yes, we will move pfn_to_online_page() > > > > > > into soft offline code. It seems that we can also move pfn_valid(), > > > > > > but is simply moving like below good enough for you? > > > > > > > > > > At least I can't am the patch to current next/master (due to > > > > > pfn_to_online_page()). > > > > > > Could also be that my "git am" skills failed as the mail was not a > > > proper patch itself :) > > > > Sorry for the inconvenience, my company email system breaks original > > message by introducing quoted-printable format ('=20' or '=3D'). > > Most mail client usually handles it but git-am doesn't. > > I give up using it and send via smtp.gmail.com. > > > > > > @@ -1877,11 +1877,17 @@ static int soft_offline_free_page(struct page *page) > > > > * This is not a 100% solution for all memory, but tries to be > > > > * ``good enough'' for the majority of memory. > > > > */ > > > > -int soft_offline_page(struct page *page, int flags) > > > > +int soft_offline_page(unsigned long pfn, int flags) > > > > { > > > > int ret; > > > > - unsigned long pfn = page_to_pfn(page); > > > > + struct page *page; > > > > + if (!pfn_valid(pfn)) > > > > + return -ENXIO; > > > > + /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ > > > > + page = pfn_to_online_page(pfn); > > > > + if (!page) > > > > + return -EIO; > > > > if (is_zone_device_page(page)) { > > > > > > -> this is now no longer possible! So you can drop the whole if > > > (is_zone_device....) case > > > > OK, thanks. I updated it. > > > > Thanks, > > Naoya Horiguchi > > --- > > From 5faf227839b578726fe7f5ff414a153abb3b3a31 Mon Sep 17 00:00:00 2001 > > From: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> > > Date: Thu, 17 Oct 2019 08:40:53 +0900 > > Subject: [PATCH] mm, soft-offline: convert parameter to pfn > > > > Currently soft_offline_page() receives struct page, and its sibling > > memory_failure() receives pfn. This discrepancy looks weird and makes > > precheck on pfn validity tricky. So let's align them. > > > > Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> > > --- > > drivers/base/memory.c | 7 +------ > > include/linux/mm.h | 2 +- > > mm/madvise.c | 2 +- > > mm/memory-failure.c | 19 +++++++++---------- > > 4 files changed, 12 insertions(+), 18 deletions(-) > > > > diff --git a/drivers/base/memory.c b/drivers/base/memory.c > > index 55907c27075b..a757d9ed88a7 100644 > > --- a/drivers/base/memory.c > > +++ b/drivers/base/memory.c > > @@ -538,12 +538,7 @@ static ssize_t soft_offline_page_store(struct device *dev, > > if (kstrtoull(buf, 0, &pfn) < 0) > > return -EINVAL; > > pfn >>= PAGE_SHIFT; > > - if (!pfn_valid(pfn)) > > - return -ENXIO; > > - /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ > > - if (!pfn_to_online_page(pfn)) > > - return -EIO; > > - ret = soft_offline_page(pfn_to_page(pfn), 0); > > + ret = soft_offline_page(pfn, 0); > > return ret == 0 ? count : ret; > > } > > diff --git a/include/linux/mm.h b/include/linux/mm.h > > index 44d058723db9..fd360d208346 100644 > > --- a/include/linux/mm.h > > +++ b/include/linux/mm.h > > @@ -2794,7 +2794,7 @@ extern int sysctl_memory_failure_early_kill; > > extern int sysctl_memory_failure_recovery; > > extern void shake_page(struct page *p, int access); > > extern atomic_long_t num_poisoned_pages __read_mostly; > > -extern int soft_offline_page(struct page *page, int flags); > > +extern int soft_offline_page(unsigned long pfn, int flags); > > /* > > diff --git a/mm/madvise.c b/mm/madvise.c > > index 2be9f3fdb05e..99dd06fecfa9 100644 > > --- a/mm/madvise.c > > +++ b/mm/madvise.c > > @@ -887,7 +887,7 @@ static int madvise_inject_error(int behavior, > > pr_info("Soft offlining pfn %#lx at process virtual address %#lx\n", > > pfn, start); > > - ret = soft_offline_page(page, MF_COUNT_INCREASED); > > + ret = soft_offline_page(pfn, MF_COUNT_INCREASED); > > if (ret) > > return ret; > > continue; > > diff --git a/mm/memory-failure.c b/mm/memory-failure.c > > index 05c8c6df25e6..af2712004a4d 100644 > > --- a/mm/memory-failure.c > > +++ b/mm/memory-failure.c > > @@ -1476,7 +1476,7 @@ static void memory_failure_work_func(struct work_struct *work) > > if (!gotten) > > break; > > if (entry.flags & MF_SOFT_OFFLINE) > > - soft_offline_page(pfn_to_page(entry.pfn), entry.flags); > > + soft_offline_page(entry.pfn, entry.flags); > > else > > memory_failure(entry.pfn, entry.flags); > > } > > @@ -1857,7 +1857,7 @@ static int soft_offline_free_page(struct page *page) > > /** > > * soft_offline_page - Soft offline a page. > > - * @page: page to offline > > + * @pfn: pfn to soft-offline > > * @flags: flags. Same as memory_failure(). > > * > > * Returns 0 on success, otherwise negated errno. > > @@ -1877,18 +1877,17 @@ static int soft_offline_free_page(struct page *page) > > * This is not a 100% solution for all memory, but tries to be > > * ``good enough'' for the majority of memory. > > */ > > -int soft_offline_page(struct page *page, int flags) > > +int soft_offline_page(unsigned long pfn, int flags) > > { > > int ret; > > - unsigned long pfn = page_to_pfn(page); > > + struct page *page; > > - if (is_zone_device_page(page)) { > > - pr_debug_ratelimited("soft_offline: %#lx page is device page\n", > > - pfn); > > - if (flags & MF_COUNT_INCREASED) > > - put_page(page); > > + if (!pfn_valid(pfn)) > > + return -ENXIO; > > + /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ > > + page = pfn_to_online_page(pfn); > > + if (!page) > > If you pass in a PFN with MF_COUNT_INCREASED via mm/madvise.c, you would now > no longer do a put_page(page) in case of ZONE_DEVICE (!page = > pfn_to_online_page(pfn);) Yes, right. > > something like this > > page = pfn_to_online_page(pfn); > if (!page) { > /* > * With MF_COUNT_INCREASED, we can use pfn_to_page() directly > * (esp., ZONE_DEVICE). > */ > if (flags & MF_COUNT_INCREASED) > put_page(pfn_to_page(page)); > return -EIO; > } > > For !pfn_valid(pfn), this is not relevant. Actually I guess that !pfn_valid() never happens when called from madvise_inject_error(), because madvise_inject_error() gets pfn via get_user_pages_fast() which only returns valid page for valid pfn. And we plan to remove MF_COUNT_INCREASED by Oscar's re-design work, so I start feeling that this patch should come on top of his tree. Thanks, Naoya Horiguchi
On 17.10.19 09:50, Naoya Horiguchi wrote: > On Thu, Oct 17, 2019 at 09:16:42AM +0200, David Hildenbrand wrote: >> On 17.10.19 01:47, Naoya Horiguchi wrote: >>> On Wed, Oct 16, 2019 at 10:57:57AM +0200, David Hildenbrand wrote: >>>> On 16.10.19 10:54, Naoya Horiguchi wrote: >>>>> On Wed, Oct 16, 2019 at 10:34:52AM +0200, David Hildenbrand wrote: >>>>>> On 16.10.19 10:27, Naoya Horiguchi wrote: >>>>>>> On Wed, Oct 16, 2019 at 09:56:19AM +0200, David Hildenbrand wrote: >>>>>>>> On 16.10.19 09:09, Naoya Horiguchi wrote: >>>>>>>>> Hi, >>>>>>>>> >>>>>>>>> I wrote a simple cleanup for parameter of soft_offline_page(), >>>>>>>>> based on thread https://lkml.org/lkml/2019/10/11/57. >>>>>>>>> >>>>>>>>> I know that we need more cleanup on hwpoison-inject, but I think >>>>>>>>> that will be mentioned in re-write patchset Oscar is preparing now. >>>>>>>>> So let me shared only this part as a separate one now. >>>>>>> ... >>>>>>>> >>>>>>>> I think you should rebase that patch on linux-next (where the >>>>>>>> pfn_to_online_page() check is in place). I assume you'll want to move the >>>>>>>> pfn_to_online_page() check into soft_offline_page() then as well? >>>>>>> >>>>>>> I rebased to next-20191016. And yes, we will move pfn_to_online_page() >>>>>>> into soft offline code. It seems that we can also move pfn_valid(), >>>>>>> but is simply moving like below good enough for you? >>>>>> >>>>>> At least I can't am the patch to current next/master (due to >>>>>> pfn_to_online_page()). >>>> >>>> Could also be that my "git am" skills failed as the mail was not a >>>> proper patch itself :) >>> >>> Sorry for the inconvenience, my company email system breaks original >>> message by introducing quoted-printable format ('=20' or '=3D'). >>> Most mail client usually handles it but git-am doesn't. >>> I give up using it and send via smtp.gmail.com. >>> >>>>> @@ -1877,11 +1877,17 @@ static int soft_offline_free_page(struct page *page) >>>>> * This is not a 100% solution for all memory, but tries to be >>>>> * ``good enough'' for the majority of memory. >>>>> */ >>>>> -int soft_offline_page(struct page *page, int flags) >>>>> +int soft_offline_page(unsigned long pfn, int flags) >>>>> { >>>>> int ret; >>>>> - unsigned long pfn = page_to_pfn(page); >>>>> + struct page *page; >>>>> + if (!pfn_valid(pfn)) >>>>> + return -ENXIO; >>>>> + /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ >>>>> + page = pfn_to_online_page(pfn); >>>>> + if (!page) >>>>> + return -EIO; >>>>> if (is_zone_device_page(page)) { >>>> >>>> -> this is now no longer possible! So you can drop the whole if >>>> (is_zone_device....) case >>> >>> OK, thanks. I updated it. >>> >>> Thanks, >>> Naoya Horiguchi >>> --- >>> From 5faf227839b578726fe7f5ff414a153abb3b3a31 Mon Sep 17 00:00:00 2001 >>> From: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> >>> Date: Thu, 17 Oct 2019 08:40:53 +0900 >>> Subject: [PATCH] mm, soft-offline: convert parameter to pfn >>> >>> Currently soft_offline_page() receives struct page, and its sibling >>> memory_failure() receives pfn. This discrepancy looks weird and makes >>> precheck on pfn validity tricky. So let's align them. >>> >>> Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> >>> --- >>> drivers/base/memory.c | 7 +------ >>> include/linux/mm.h | 2 +- >>> mm/madvise.c | 2 +- >>> mm/memory-failure.c | 19 +++++++++---------- >>> 4 files changed, 12 insertions(+), 18 deletions(-) >>> >>> diff --git a/drivers/base/memory.c b/drivers/base/memory.c >>> index 55907c27075b..a757d9ed88a7 100644 >>> --- a/drivers/base/memory.c >>> +++ b/drivers/base/memory.c >>> @@ -538,12 +538,7 @@ static ssize_t soft_offline_page_store(struct device *dev, >>> if (kstrtoull(buf, 0, &pfn) < 0) >>> return -EINVAL; >>> pfn >>= PAGE_SHIFT; >>> - if (!pfn_valid(pfn)) >>> - return -ENXIO; >>> - /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ >>> - if (!pfn_to_online_page(pfn)) >>> - return -EIO; >>> - ret = soft_offline_page(pfn_to_page(pfn), 0); >>> + ret = soft_offline_page(pfn, 0); >>> return ret == 0 ? count : ret; >>> } >>> diff --git a/include/linux/mm.h b/include/linux/mm.h >>> index 44d058723db9..fd360d208346 100644 >>> --- a/include/linux/mm.h >>> +++ b/include/linux/mm.h >>> @@ -2794,7 +2794,7 @@ extern int sysctl_memory_failure_early_kill; >>> extern int sysctl_memory_failure_recovery; >>> extern void shake_page(struct page *p, int access); >>> extern atomic_long_t num_poisoned_pages __read_mostly; >>> -extern int soft_offline_page(struct page *page, int flags); >>> +extern int soft_offline_page(unsigned long pfn, int flags); >>> /* >>> diff --git a/mm/madvise.c b/mm/madvise.c >>> index 2be9f3fdb05e..99dd06fecfa9 100644 >>> --- a/mm/madvise.c >>> +++ b/mm/madvise.c >>> @@ -887,7 +887,7 @@ static int madvise_inject_error(int behavior, >>> pr_info("Soft offlining pfn %#lx at process virtual address %#lx\n", >>> pfn, start); >>> - ret = soft_offline_page(page, MF_COUNT_INCREASED); >>> + ret = soft_offline_page(pfn, MF_COUNT_INCREASED); >>> if (ret) >>> return ret; >>> continue; >>> diff --git a/mm/memory-failure.c b/mm/memory-failure.c >>> index 05c8c6df25e6..af2712004a4d 100644 >>> --- a/mm/memory-failure.c >>> +++ b/mm/memory-failure.c >>> @@ -1476,7 +1476,7 @@ static void memory_failure_work_func(struct work_struct *work) >>> if (!gotten) >>> break; >>> if (entry.flags & MF_SOFT_OFFLINE) >>> - soft_offline_page(pfn_to_page(entry.pfn), entry.flags); >>> + soft_offline_page(entry.pfn, entry.flags); >>> else >>> memory_failure(entry.pfn, entry.flags); >>> } >>> @@ -1857,7 +1857,7 @@ static int soft_offline_free_page(struct page *page) >>> /** >>> * soft_offline_page - Soft offline a page. >>> - * @page: page to offline >>> + * @pfn: pfn to soft-offline >>> * @flags: flags. Same as memory_failure(). >>> * >>> * Returns 0 on success, otherwise negated errno. >>> @@ -1877,18 +1877,17 @@ static int soft_offline_free_page(struct page *page) >>> * This is not a 100% solution for all memory, but tries to be >>> * ``good enough'' for the majority of memory. >>> */ >>> -int soft_offline_page(struct page *page, int flags) >>> +int soft_offline_page(unsigned long pfn, int flags) >>> { >>> int ret; >>> - unsigned long pfn = page_to_pfn(page); >>> + struct page *page; >>> - if (is_zone_device_page(page)) { >>> - pr_debug_ratelimited("soft_offline: %#lx page is device page\n", >>> - pfn); >>> - if (flags & MF_COUNT_INCREASED) >>> - put_page(page); >>> + if (!pfn_valid(pfn)) >>> + return -ENXIO; >>> + /* Only online pages can be soft-offlined (esp., not ZONE_DEVICE). */ >>> + page = pfn_to_online_page(pfn); >>> + if (!page) >> >> If you pass in a PFN with MF_COUNT_INCREASED via mm/madvise.c, you would now >> no longer do a put_page(page) in case of ZONE_DEVICE (!page = >> pfn_to_online_page(pfn);) > > Yes, right. > >> >> something like this >> >> page = pfn_to_online_page(pfn); >> if (!page) { >> /* >> * With MF_COUNT_INCREASED, we can use pfn_to_page() directly >> * (esp., ZONE_DEVICE). >> */ >> if (flags & MF_COUNT_INCREASED) >> put_page(pfn_to_page(page)); >> return -EIO; >> } >> >> For !pfn_valid(pfn), this is not relevant. > > Actually I guess that !pfn_valid() never happens when called from > madvise_inject_error(), because madvise_inject_error() gets pfn via > get_user_pages_fast() which only returns valid page for valid pfn. > > And we plan to remove MF_COUNT_INCREASED by Oscar's re-design work, > so I start feeling that this patch should come on top of his tree. I agree, let's clean that up first.
On Thu, Oct 17, 2019 at 07:50:18AM +0000, Naoya Horiguchi wrote: > Actually I guess that !pfn_valid() never happens when called from > madvise_inject_error(), because madvise_inject_error() gets pfn via > get_user_pages_fast() which only returns valid page for valid pfn. > > And we plan to remove MF_COUNT_INCREASED by Oscar's re-design work, > so I start feeling that this patch should come on top of his tree. Hi Naoya, I am pretty much done with my testing. If you feel like, I can take the patch and add it on top of [1] , then I will do some more testing and, if nothing pops up, I will send it upstream. [1] https://github.com/leberus/linux/tree/hwpoison-v2 > > Thanks, > Naoya Horiguchi
On Thu, Oct 17, 2019 at 10:03:21AM +0200, Oscar Salvador wrote: > On Thu, Oct 17, 2019 at 07:50:18AM +0000, Naoya Horiguchi wrote: > > Actually I guess that !pfn_valid() never happens when called from > > madvise_inject_error(), because madvise_inject_error() gets pfn via > > get_user_pages_fast() which only returns valid page for valid pfn. > > > > And we plan to remove MF_COUNT_INCREASED by Oscar's re-design work, > > so I start feeling that this patch should come on top of his tree. > > Hi Naoya, > > I am pretty much done with my testing. > If you feel like, I can take the patch and add it on top of [1] > , then I will do some more testing and, if nothing pops up, I will > send it upstream. Hi Oscar, Yes, please take it, thank you for speaking up. > > [1] https://github.com/leberus/linux/tree/hwpoison-v2
diff --git a/drivers/base/memory.c b/drivers/base/memory.c index e5485c22ef77..04e469c82852 100644 --- a/drivers/base/memory.c +++ b/drivers/base/memory.c @@ -540,7 +540,7 @@ static ssize_t soft_offline_page_store(struct device *dev, pfn >>= PAGE_SHIFT; if (!pfn_valid(pfn)) return -ENXIO; - ret = soft_offline_page(pfn_to_page(pfn)); + ret = soft_offline_page(pfn); return ret == 0 ? count : ret; } diff --git a/include/linux/mm.h b/include/linux/mm.h index 3eba26324ff1..0a452020edf5 100644 --- a/include/linux/mm.h +++ b/include/linux/mm.h @@ -2783,7 +2783,7 @@ extern int sysctl_memory_failure_early_kill; extern int sysctl_memory_failure_recovery; extern void shake_page(struct page *p, int access); extern atomic_long_t num_poisoned_pages __read_mostly; -extern int soft_offline_page(struct page *page); +extern int soft_offline_page(unsigned long pfn); /* diff --git a/mm/madvise.c b/mm/madvise.c index fd221b610b52..df198d1e5e2e 100644 --- a/mm/madvise.c +++ b/mm/madvise.c @@ -893,7 +893,7 @@ static int madvise_inject_error(int behavior, if (behavior == MADV_SOFT_OFFLINE) { pr_info("Soft offlining pfn %#lx at process virtual address %#lx\n", pfn, start); - ret = soft_offline_page(page); + ret = soft_offline_page(pfn); if (ret) return ret; } else { diff --git a/mm/memory-failure.c b/mm/memory-failure.c index 4f16e0a7e7cc..eb4fd5e8d5e1 100644 --- a/mm/memory-failure.c +++ b/mm/memory-failure.c @@ -1514,7 +1514,7 @@ static void memory_failure_work_func(struct work_struct *work) if (!gotten) break; if (entry.flags & MF_SOFT_OFFLINE) - soft_offline_page(pfn_to_page(entry.pfn)); + soft_offline_page(entry.pfn); else memory_failure(entry.pfn, entry.flags); } @@ -1822,7 +1822,7 @@ static int soft_offline_free_page(struct page *page) /** * soft_offline_page - Soft offline a page. - * @page: page to offline + * @pfn: pfn to soft-offline * * Returns 0 on success, otherwise negated errno. * @@ -1841,10 +1841,10 @@ static int soft_offline_free_page(struct page *page) * This is not a 100% solution for all memory, but tries to be * ``good enough'' for the majority of memory. */ -int soft_offline_page(struct page *page) +int soft_offline_page(unsigned long pfn) { int ret; - unsigned long pfn = page_to_pfn(page); + struct page *page = pfn_to_page(pfn); if (is_zone_device_page(page)) { pr_debug_ratelimited("soft_offline: %#lx page is device page\n",