* v5.12+ regression on page_poison=1 i_on_free=0 i_on_alloc=0 warnings @ 2021-07-11 23:57 Sergei Trofimovich 2021-07-12 21:58 ` [PATCH] mm: page_alloc: fix page_poison=1 / INIT_ON_ALLOC_DEFAULT_ON interaction Sergei Trofimovich 0 siblings, 1 reply; 7+ messages in thread From: Sergei Trofimovich @ 2021-07-11 23:57 UTC (permalink / raw) To: linux-mm, Kees Cook Cc: Thomas Gleixner, Alexander Potapenko, Vlastimil Babka, Andrey Konovalov, bowsingbetee In https://bugs.gentoo.org/801481 https://bugzilla.kernel.org/show_bug.cgi?id=213697 bowsingbetee noticed that the following setup causes kernel to complain about poisoning on memset(0) pages: - kernel command: page_poison=1 init_on_free=0 init_on_alloc=0 - kernel config: * CONFIG_INIT_ON_ALLOC_DEFAULT_ON=y * CONFIG_INIT_ON_FREE_DEFAULT_ON=y * CONFIG_PAGE_POISONING=y v5.12 works ok, boots as: [ 0.009691][ T0] mem auto-init: stack:off, heap alloc:off, heap free:off v5.13 warns, boots as: [ 0.009746][ T0] mem auto-init: stack:off, heap alloc:on, heap free:on Current 5.14-rc1 is also affected: [ 0.009233][ T0] mem auto-init: stack:off, heap alloc:on, heap free:on [ 0.009234][ T0] mem auto-init: clearing system memory may take some time... [ 0.396784][ T0] pagealloc: memory corruption [ 0.396788][ T0] ffff888100000000: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ I attempted to bisect it and found the follwing relevant commits: Before the following commit no warnings were present: commit 51cba1ebc60df9c4ce034a9f5441169c0d0956c0 Author: Kees Cook <keescook@chromium.org> Date: Thu Apr 1 16:23:43 2021 -0700 init_on_alloc: Optimize static branches Next after this commit kernel does not boot. Makes it harder to bisect the poisoning warning After the commit below kernel boots and already complains about poisoning of zero pages: commit 1bb5eab30d68c1a3d9dbc822e1895e6c06dbe748 Author: Andrey Konovalov <andreyknvl@google.com> Date: Thu Apr 29 23:00:02 2021 -0700 kasan, mm: integrate page_alloc init with HW_TAGS It looks like a case of page_poison=1 not having enough of a priority over init_on_free=0 init_on_alloc=0 at early boot. -- Sergei ^ permalink raw reply [flat|nested] 7+ messages in thread
* [PATCH] mm: page_alloc: fix page_poison=1 / INIT_ON_ALLOC_DEFAULT_ON interaction 2021-07-11 23:57 v5.12+ regression on page_poison=1 i_on_free=0 i_on_alloc=0 warnings Sergei Trofimovich @ 2021-07-12 21:58 ` Sergei Trofimovich 2021-07-14 2:00 ` Andrew Morton 2021-07-14 3:10 ` Kees Cook 0 siblings, 2 replies; 7+ messages in thread From: Sergei Trofimovich @ 2021-07-12 21:58 UTC (permalink / raw) To: linux-mm Cc: linux-kernel, Sergei Trofimovich, Andrew Morton, Kees Cook, Alexander Potapenko, Thomas Gleixner, Vlastimil Babka, bowsingbetee To reproduce the failure we need the following system: - kernel command: page_poison=1 init_on_free=0 init_on_alloc=0 - kernel config: * CONFIG_INIT_ON_ALLOC_DEFAULT_ON=y * CONFIG_INIT_ON_FREE_DEFAULT_ON=y * CONFIG_PAGE_POISONING=y 0000000085629bdd: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ 0000000022861832: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ 00000000c597f5b0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ CPU: 11 PID: 15195 Comm: bash Kdump: loaded Tainted: G U O 5.13.1-gentoo-x86_64 #1 Hardware name: System manufacturer System Product Name/PRIME Z370-A, BIOS 2801 01/13/2021 Call Trace: dump_stack+0x64/0x7c __kernel_unpoison_pages.cold+0x48/0x84 post_alloc_hook+0x60/0xa0 get_page_from_freelist+0xdb8/0x1000 __alloc_pages+0x163/0x2b0 __get_free_pages+0xc/0x30 pgd_alloc+0x2e/0x1a0 ? dup_mm+0x37/0x4f0 mm_init+0x185/0x270 dup_mm+0x6b/0x4f0 ? __lock_task_sighand+0x35/0x70 copy_process+0x190d/0x1b10 kernel_clone+0xba/0x3b0 __do_sys_clone+0x8f/0xb0 do_syscall_64+0x68/0x80 ? do_syscall_64+0x11/0x80 entry_SYSCALL_64_after_hwframe+0x44/0xae Before the 51cba1eb ("init_on_alloc: Optimize static branches") init_on_alloc never enabled static branch by default. It could only be enabed explicitly by init_mem_debugging_and_hardening(). But after the 51cba1eb static branch could already be enabled by default. There was no code to ever disable it. That caused page_poison=1 / init_on_free=1 conflict. This change extends init_mem_debugging_and_hardening() to also disable static branch disabling. CC: Andrew Morton <akpm@linux-foundation.org> CC: Kees Cook <keescook@chromium.org> CC: Alexander Potapenko <glider@google.com> CC: Thomas Gleixner <tglx@linutronix.de> CC: Vlastimil Babka <vbabka@suse.cz> CC: linux-mm@kvack.org Reported-by: bowsingbetee@pm.me Reported-by: Mikhail Morfikov Fixes: 51cba1eb ("init_on_alloc: Optimize static branches") Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org> --- mm/page_alloc.c | 16 ++++++++++------ 1 file changed, 10 insertions(+), 6 deletions(-) diff --git a/mm/page_alloc.c b/mm/page_alloc.c index 3b97e17806be..46cb4a9c2b50 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -840,18 +840,22 @@ void init_mem_debugging_and_hardening(void) } #endif - if (_init_on_alloc_enabled_early) { - if (page_poisoning_requested) + if (_init_on_alloc_enabled_early || + IS_ENABLED(CONFIG_INIT_ON_ALLOC_DEFAULT_ON)) { + if (page_poisoning_requested) { pr_info("mem auto-init: CONFIG_PAGE_POISONING is on, " "will take precedence over init_on_alloc\n"); - else + static_branch_disable(&init_on_alloc); + } else static_branch_enable(&init_on_alloc); } - if (_init_on_free_enabled_early) { - if (page_poisoning_requested) + if (_init_on_free_enabled_early || + IS_ENABLED(CONFIG_INIT_ON_FREE_DEFAULT_ON)) { + if (page_poisoning_requested) { pr_info("mem auto-init: CONFIG_PAGE_POISONING is on, " "will take precedence over init_on_free\n"); - else + static_branch_disable(&init_on_free); + } else static_branch_enable(&init_on_free); } -- 2.32.0 ^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH] mm: page_alloc: fix page_poison=1 / INIT_ON_ALLOC_DEFAULT_ON interaction 2021-07-12 21:58 ` [PATCH] mm: page_alloc: fix page_poison=1 / INIT_ON_ALLOC_DEFAULT_ON interaction Sergei Trofimovich @ 2021-07-14 2:00 ` Andrew Morton 2021-07-17 18:18 ` Sergei Trofimovich 2021-07-14 3:10 ` Kees Cook 1 sibling, 1 reply; 7+ messages in thread From: Andrew Morton @ 2021-07-14 2:00 UTC (permalink / raw) To: Sergei Trofimovich Cc: linux-mm, linux-kernel, Kees Cook, Alexander Potapenko, Thomas Gleixner, Vlastimil Babka, bowsingbetee On Mon, 12 Jul 2021 22:58:16 +0100 Sergei Trofimovich <slyfox@gentoo.org> wrote: > To reproduce the failure we need the following system: > - kernel command: page_poison=1 init_on_free=0 init_on_alloc=0 > - kernel config: > * CONFIG_INIT_ON_ALLOC_DEFAULT_ON=y > * CONFIG_INIT_ON_FREE_DEFAULT_ON=y > * CONFIG_PAGE_POISONING=y > > 0000000085629bdd: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ > 0000000022861832: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ > 00000000c597f5b0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ > CPU: 11 PID: 15195 Comm: bash Kdump: loaded Tainted: G U O 5.13.1-gentoo-x86_64 #1 > Hardware name: System manufacturer System Product Name/PRIME Z370-A, BIOS 2801 01/13/2021 > Call Trace: > dump_stack+0x64/0x7c > __kernel_unpoison_pages.cold+0x48/0x84 > post_alloc_hook+0x60/0xa0 > get_page_from_freelist+0xdb8/0x1000 > __alloc_pages+0x163/0x2b0 > __get_free_pages+0xc/0x30 > pgd_alloc+0x2e/0x1a0 > ? dup_mm+0x37/0x4f0 > mm_init+0x185/0x270 > dup_mm+0x6b/0x4f0 > ? __lock_task_sighand+0x35/0x70 > copy_process+0x190d/0x1b10 > kernel_clone+0xba/0x3b0 > __do_sys_clone+0x8f/0xb0 > do_syscall_64+0x68/0x80 > ? do_syscall_64+0x11/0x80 > entry_SYSCALL_64_after_hwframe+0x44/0xae > > Before the 51cba1eb ("init_on_alloc: Optimize static branches") > init_on_alloc never enabled static branch by default. It could > only be enabed explicitly by init_mem_debugging_and_hardening(). > > But after the 51cba1eb static branch could already be enabled > by default. There was no code to ever disable it. That caused > page_poison=1 / init_on_free=1 conflict. > > This change extends init_mem_debugging_and_hardening() to also > disable static branch disabling. > > ... > > --- a/mm/page_alloc.c > +++ b/mm/page_alloc.c > @@ -840,18 +840,22 @@ void init_mem_debugging_and_hardening(void) > } > #endif > > - if (_init_on_alloc_enabled_early) { > - if (page_poisoning_requested) > + if (_init_on_alloc_enabled_early || > + IS_ENABLED(CONFIG_INIT_ON_ALLOC_DEFAULT_ON)) { > + if (page_poisoning_requested) { > pr_info("mem auto-init: CONFIG_PAGE_POISONING is on, " > "will take precedence over init_on_alloc\n"); > - else > + static_branch_disable(&init_on_alloc); > + } else > static_branch_enable(&init_on_alloc); > } > - if (_init_on_free_enabled_early) { > - if (page_poisoning_requested) > + if (_init_on_free_enabled_early || > + IS_ENABLED(CONFIG_INIT_ON_FREE_DEFAULT_ON)) { > + if (page_poisoning_requested) { > pr_info("mem auto-init: CONFIG_PAGE_POISONING is on, " > "will take precedence over init_on_free\n"); > - else > + static_branch_disable(&init_on_free); > + } else > static_branch_enable(&init_on_free); > } > I'm thinking this is sufficiently serious and sufficiently reported to warrant a cc:stable backport. Agree? ^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH] mm: page_alloc: fix page_poison=1 / INIT_ON_ALLOC_DEFAULT_ON interaction 2021-07-14 2:00 ` Andrew Morton @ 2021-07-17 18:18 ` Sergei Trofimovich 0 siblings, 0 replies; 7+ messages in thread From: Sergei Trofimovich @ 2021-07-17 18:18 UTC (permalink / raw) To: Andrew Morton Cc: linux-mm, linux-kernel, Kees Cook, Alexander Potapenko, Thomas Gleixner, Vlastimil Babka, bowsingbetee On Tue, 13 Jul 2021 19:00:51 -0700 Andrew Morton <akpm@linux-foundation.org> wrote: > On Mon, 12 Jul 2021 22:58:16 +0100 Sergei Trofimovich <slyfox@gentoo.org> wrote: > > > To reproduce the failure we need the following system: > > - kernel command: page_poison=1 init_on_free=0 init_on_alloc=0 > > - kernel config: > > * CONFIG_INIT_ON_ALLOC_DEFAULT_ON=y > > * CONFIG_INIT_ON_FREE_DEFAULT_ON=y > > * CONFIG_PAGE_POISONING=y > > > > 0000000085629bdd: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ > > 0000000022861832: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ > > 00000000c597f5b0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ > > CPU: 11 PID: 15195 Comm: bash Kdump: loaded Tainted: G U O 5.13.1-gentoo-x86_64 #1 > > Hardware name: System manufacturer System Product Name/PRIME Z370-A, BIOS 2801 01/13/2021 > > Call Trace: > > dump_stack+0x64/0x7c > > __kernel_unpoison_pages.cold+0x48/0x84 > > post_alloc_hook+0x60/0xa0 > > get_page_from_freelist+0xdb8/0x1000 > > __alloc_pages+0x163/0x2b0 > > __get_free_pages+0xc/0x30 > > pgd_alloc+0x2e/0x1a0 > > ? dup_mm+0x37/0x4f0 > > mm_init+0x185/0x270 > > dup_mm+0x6b/0x4f0 > > ? __lock_task_sighand+0x35/0x70 > > copy_process+0x190d/0x1b10 > > kernel_clone+0xba/0x3b0 > > __do_sys_clone+0x8f/0xb0 > > do_syscall_64+0x68/0x80 > > ? do_syscall_64+0x11/0x80 > > entry_SYSCALL_64_after_hwframe+0x44/0xae > > > > Before the 51cba1eb ("init_on_alloc: Optimize static branches") > > init_on_alloc never enabled static branch by default. It could > > only be enabed explicitly by init_mem_debugging_and_hardening(). > > > > But after the 51cba1eb static branch could already be enabled > > by default. There was no code to ever disable it. That caused > > page_poison=1 / init_on_free=1 conflict. > > > > This change extends init_mem_debugging_and_hardening() to also > > disable static branch disabling. > > > > ... > > > > --- a/mm/page_alloc.c > > +++ b/mm/page_alloc.c > > @@ -840,18 +840,22 @@ void init_mem_debugging_and_hardening(void) > > } > > #endif > > > > - if (_init_on_alloc_enabled_early) { > > - if (page_poisoning_requested) > > + if (_init_on_alloc_enabled_early || > > + IS_ENABLED(CONFIG_INIT_ON_ALLOC_DEFAULT_ON)) { > > + if (page_poisoning_requested) { > > pr_info("mem auto-init: CONFIG_PAGE_POISONING is on, " > > "will take precedence over init_on_alloc\n"); > > - else > > + static_branch_disable(&init_on_alloc); > > + } else > > static_branch_enable(&init_on_alloc); > > } > > - if (_init_on_free_enabled_early) { > > - if (page_poisoning_requested) > > + if (_init_on_free_enabled_early || > > + IS_ENABLED(CONFIG_INIT_ON_FREE_DEFAULT_ON)) { > > + if (page_poisoning_requested) { > > pr_info("mem auto-init: CONFIG_PAGE_POISONING is on, " > > "will take precedence over init_on_free\n"); > > - else > > + static_branch_disable(&init_on_free); > > + } else > > static_branch_enable(&init_on_free); > > } > > > > I'm thinking this is sufficiently serious and sufficiently reported to > warrant a cc:stable backport. Agree? I agree. The patch might be tricky to apply as is too far back. But current release should be fine. -- Sergei ^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH] mm: page_alloc: fix page_poison=1 / INIT_ON_ALLOC_DEFAULT_ON interaction 2021-07-12 21:58 ` [PATCH] mm: page_alloc: fix page_poison=1 / INIT_ON_ALLOC_DEFAULT_ON interaction Sergei Trofimovich 2021-07-14 2:00 ` Andrew Morton @ 2021-07-14 3:10 ` Kees Cook 2021-07-14 11:11 ` bowsingbetee 1 sibling, 1 reply; 7+ messages in thread From: Kees Cook @ 2021-07-14 3:10 UTC (permalink / raw) To: Sergei Trofimovich Cc: linux-mm, linux-kernel, Andrew Morton, Alexander Potapenko, Thomas Gleixner, Vlastimil Babka, bowsingbetee On Mon, Jul 12, 2021 at 10:58:16PM +0100, Sergei Trofimovich wrote: > To reproduce the failure we need the following system: > - kernel command: page_poison=1 init_on_free=0 init_on_alloc=0 > - kernel config: > * CONFIG_INIT_ON_ALLOC_DEFAULT_ON=y > * CONFIG_INIT_ON_FREE_DEFAULT_ON=y > * CONFIG_PAGE_POISONING=y > > 0000000085629bdd: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ > 0000000022861832: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ > 00000000c597f5b0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ > CPU: 11 PID: 15195 Comm: bash Kdump: loaded Tainted: G U O 5.13.1-gentoo-x86_64 #1 > Hardware name: System manufacturer System Product Name/PRIME Z370-A, BIOS 2801 01/13/2021 > Call Trace: > dump_stack+0x64/0x7c > __kernel_unpoison_pages.cold+0x48/0x84 > post_alloc_hook+0x60/0xa0 > get_page_from_freelist+0xdb8/0x1000 > __alloc_pages+0x163/0x2b0 > __get_free_pages+0xc/0x30 > pgd_alloc+0x2e/0x1a0 > ? dup_mm+0x37/0x4f0 > mm_init+0x185/0x270 > dup_mm+0x6b/0x4f0 > ? __lock_task_sighand+0x35/0x70 > copy_process+0x190d/0x1b10 > kernel_clone+0xba/0x3b0 > __do_sys_clone+0x8f/0xb0 > do_syscall_64+0x68/0x80 > ? do_syscall_64+0x11/0x80 > entry_SYSCALL_64_after_hwframe+0x44/0xae > > Before the 51cba1eb ("init_on_alloc: Optimize static branches") > init_on_alloc never enabled static branch by default. It could > only be enabed explicitly by init_mem_debugging_and_hardening(). But init_mem_debugging_and_hardening() is always called (by mm_init()). > But after the 51cba1eb static branch could already be enabled > by default. There was no code to ever disable it. That caused > page_poison=1 / init_on_free=1 conflict. > > This change extends init_mem_debugging_and_hardening() to also > disable static branch disabling. > > CC: Andrew Morton <akpm@linux-foundation.org> > CC: Kees Cook <keescook@chromium.org> > CC: Alexander Potapenko <glider@google.com> > CC: Thomas Gleixner <tglx@linutronix.de> > CC: Vlastimil Babka <vbabka@suse.cz> > CC: linux-mm@kvack.org > Reported-by: bowsingbetee@pm.me > Reported-by: Mikhail Morfikov > Fixes: 51cba1eb ("init_on_alloc: Optimize static branches") > Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org> > --- > mm/page_alloc.c | 16 ++++++++++------ > 1 file changed, 10 insertions(+), 6 deletions(-) > > diff --git a/mm/page_alloc.c b/mm/page_alloc.c > index 3b97e17806be..46cb4a9c2b50 100644 > --- a/mm/page_alloc.c > +++ b/mm/page_alloc.c > @@ -840,18 +840,22 @@ void init_mem_debugging_and_hardening(void) > } > #endif > > - if (_init_on_alloc_enabled_early) { > - if (page_poisoning_requested) > + if (_init_on_alloc_enabled_early || > + IS_ENABLED(CONFIG_INIT_ON_ALLOC_DEFAULT_ON)) { This doesn't look right. _init_on_alloc_enabled_early already has the same value: static bool _init_on_alloc_enabled_early __read_mostly = IS_ENABLED(CONFIG_INIT_ON_ALLOC_DEFAULT_ON); So checking this is just a side-effect of how static_branch_maybe() happens to be behaving. > + if (page_poisoning_requested) { > pr_info("mem auto-init: CONFIG_PAGE_POISONING is on, " > "will take precedence over init_on_alloc\n"); > - else > + static_branch_disable(&init_on_alloc); > + } else > static_branch_enable(&init_on_alloc); > } > - if (_init_on_free_enabled_early) { > - if (page_poisoning_requested) > + if (_init_on_free_enabled_early || > + IS_ENABLED(CONFIG_INIT_ON_FREE_DEFAULT_ON)) { > + if (page_poisoning_requested) { > pr_info("mem auto-init: CONFIG_PAGE_POISONING is on, " > "will take precedence over init_on_free\n"); > - else > + static_branch_disable(&init_on_free); > + } else > static_branch_enable(&init_on_free); > } I think it would be better to clean this up without additional confusion involving the CONFIGs: diff --git a/mm/page_alloc.c b/mm/page_alloc.c index 3b97e17806be..1f19365bc158 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -840,21 +840,24 @@ void init_mem_debugging_and_hardening(void) } #endif - if (_init_on_alloc_enabled_early) { - if (page_poisoning_requested) - pr_info("mem auto-init: CONFIG_PAGE_POISONING is on, " - "will take precedence over init_on_alloc\n"); - else - static_branch_enable(&init_on_alloc); - } - if (_init_on_free_enabled_early) { - if (page_poisoning_requested) - pr_info("mem auto-init: CONFIG_PAGE_POISONING is on, " - "will take precedence over init_on_free\n"); - else - static_branch_enable(&init_on_free); + if ((_init_on_alloc_enabled_early || _init_on_free_enabled_early) && + page_poisoning_requested) { + pr_info("mem auto-init: CONFIG_PAGE_POISONING is on, " + "will take precedence over init_on_alloc and init_on_free\n"); + _init_on_alloc_enabled_early = false; + _init_on_free_enabled_early = false; } + if (_init_on_alloc_enabled_early) + static_branch_enable(&init_on_alloc); + else + static_branch_disable(&init_on_alloc); + + if (_init_on_free_enabled_early) + static_branch_enable(&init_on_free); + else + static_branch_disable(&init_on_free); + #ifdef CONFIG_DEBUG_PAGEALLOC if (!debug_pagealloc_enabled()) return; -- Kees Cook ^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH] mm: page_alloc: fix page_poison=1 / INIT_ON_ALLOC_DEFAULT_ON interaction 2021-07-14 3:10 ` Kees Cook @ 2021-07-14 11:11 ` bowsingbetee 2021-07-14 21:14 ` Kees Cook 0 siblings, 1 reply; 7+ messages in thread From: bowsingbetee @ 2021-07-14 11:11 UTC (permalink / raw) To: Kees Cook Cc: Sergei Trofimovich, linux-mm, linux-kernel, Andrew Morton, Alexander Potapenko, Thomas Gleixner, Vlastimil Babka [-- Attachment #1.1: Type: text/plain, Size: 8604 bytes --] ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐ On Wednesday, July 14th, 2021 at 5:10 AM, Kees Cook <keescook@chromium.org> wrote: > On Mon, Jul 12, 2021 at 10:58:16PM +0100, Sergei Trofimovich wrote: > > > To reproduce the failure we need the following system: > > > > - kernel command: page_poison=1 init_on_free=0 init_on_alloc=0 > > > > - kernel config: > > > > - CONFIG_INIT_ON_ALLOC_DEFAULT_ON=y > > - CONFIG_INIT_ON_FREE_DEFAULT_ON=y > > - CONFIG_PAGE_POISONING=y > > > > 0000000085629bdd: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ > > > > 0000000022861832: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ > > > > 00000000c597f5b0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ > > > > CPU: 11 PID: 15195 Comm: bash Kdump: loaded Tainted: G U O 5.13.1-gentoo-x86_64 #1 > > > > Hardware name: System manufacturer System Product Name/PRIME Z370-A, BIOS 2801 01/13/2021 > > > > Call Trace: > > > > dump_stack+0x64/0x7c > > > > __kernel_unpoison_pages.cold+0x48/0x84 > > > > post_alloc_hook+0x60/0xa0 > > > > get_page_from_freelist+0xdb8/0x1000 > > > > __alloc_pages+0x163/0x2b0 > > > > __get_free_pages+0xc/0x30 > > > > pgd_alloc+0x2e/0x1a0 > > > > ? dup_mm+0x37/0x4f0 > > > > mm_init+0x185/0x270 > > > > dup_mm+0x6b/0x4f0 > > > > ? __lock_task_sighand+0x35/0x70 > > > > copy_process+0x190d/0x1b10 > > > > kernel_clone+0xba/0x3b0 > > > > __do_sys_clone+0x8f/0xb0 > > > > do_syscall_64+0x68/0x80 > > > > ? do_syscall_64+0x11/0x80 > > > > entry_SYSCALL_64_after_hwframe+0x44/0xae > > > > > > Before the 51cba1eb ("init_on_alloc: Optimize static branches") > > > > init_on_alloc never enabled static branch by default. It could > > > > only be enabed explicitly by init_mem_debugging_and_hardening(). > > But init_mem_debugging_and_hardening() is always called (by mm_init()). > > > But after the 51cba1eb static branch could already be enabled > > > > by default. There was no code to ever disable it. That caused > > > > page_poison=1 / init_on_free=1 conflict. > > > > This change extends init_mem_debugging_and_hardening() to also > > > > disable static branch disabling. > > > > CC: Andrew Morton akpm@linux-foundation.org > > > > CC: Kees Cook keescook@chromium.org > > > > CC: Alexander Potapenko glider@google.com > > > > CC: Thomas Gleixner tglx@linutronix.de > > > > CC: Vlastimil Babka vbabka@suse.cz > > > > CC: linux-mm@kvack.org > > > > Reported-by: bowsingbetee@pm.me > > > > Reported-by: Mikhail Morfikov > > > > Fixes: 51cba1eb ("init_on_alloc: Optimize static branches") > > > > Signed-off-by: Sergei Trofimovich slyfox@gentoo.org > > ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- > > > > mm/page_alloc.c | 16 ++++++++++------ > > > > 1 file changed, 10 insertions(+), 6 deletions(-) > > > > diff --git a/mm/page_alloc.c b/mm/page_alloc.c > > > > index 3b97e17806be..46cb4a9c2b50 100644 > > > > --- a/mm/page_alloc.c > > > > +++ b/mm/page_alloc.c > > > > @@ -840,18 +840,22 @@ void init_mem_debugging_and_hardening(void) > > > > } > > > > #endif > > > > - if (_init_on_alloc_enabled_early) { > > - if (page_poisoning_requested) > > > > > > > > - if (_init_on_alloc_enabled_early || > > - IS_ENABLED(CONFIG_INIT_ON_ALLOC_DEFAULT_ON)) { > > > > > > This doesn't look right. _init_on_alloc_enabled_early already has the > > same value: > > static bool _init_on_alloc_enabled_early __read_mostly > > = IS_ENABLED(CONFIG_INIT_ON_ALLOC_DEFAULT_ON); > > So checking this is just a side-effect of how static_branch_maybe() > > happens to be behaving. > > > - if (page_poisoning_requested) { > > pr_info("mem auto-init: CONFIG_PAGE_POISONING is on, " > > "will take precedence over init_on_alloc\\n"); > > > > > > > > - else > > > > > > > > - static_branch_disable(&init_on_alloc); > > > > > > - } else > > static_branch_enable(&init_on_alloc); > > > > > > } > > > > - if (_init_on_free_enabled_early) { > > - if (page_poisoning_requested) > > > > > > > > - if (_init_on_free_enabled_early || > > - IS_ENABLED(CONFIG_INIT_ON_FREE_DEFAULT_ON)) { > > > > > > - if (page_poisoning_requested) { > > pr_info("mem auto-init: CONFIG_PAGE_POISONING is on, " > > "will take precedence over init_on_free\\n"); > > > > > > > > - else > > > > > > > > - static_branch_disable(&init_on_free); > > > > > > - } else > > static_branch_enable(&init_on_free); > > > > > > } > > I think it would be better to clean this up without additional > > confusion involving the CONFIGs: > > diff --git a/mm/page_alloc.c b/mm/page_alloc.c > > index 3b97e17806be..1f19365bc158 100644 > > --- a/mm/page_alloc.c > > +++ b/mm/page_alloc.c > > @@ -840,21 +840,24 @@ void init_mem_debugging_and_hardening(void) > > } > > #endif > > - if (_init_on_alloc_enabled_early) { > - if (page_poisoning_requested) > > > - pr_info("mem auto-init: CONFIG_PAGE_POISONING is on, " > > > - "will take precedence over init_on_alloc\\n"); > > > - else > > > - static_branch_enable(&init_on_alloc); > > > - } > - if (_init_on_free_enabled_early) { > - if (page_poisoning_requested) > > > - pr_info("mem auto-init: CONFIG_PAGE_POISONING is on, " > > > - "will take precedence over init_on_free\\n"); > > > - else > > > - static_branch_enable(&init_on_free); > > > > - if ((_init_on_alloc_enabled_early || _init_on_free_enabled_early) && > - page_poisoning_requested) { > > > - pr_info("mem auto-init: CONFIG_PAGE_POISONING is on, " > > > - "will take precedence over init_on_alloc and init_on_free\\n"); > B & C variants show this message and A does not, which is what I would expect to happen. Tested variants: A. "page_poison=1 init_on_free=0 init_on_alloc=0 slub_debug=P" B. "page_poison=1 init_on_free=0 init_on_alloc=0 slub_debug=P init_on_free=1" C. "page_poison=1 slub_debug=P" in common: CONFIG_INIT_ON_ALLOC_DEFAULT_ON=y CONFIG_INIT_ON_FREE_DEFAULT_ON=y CONFIG_PAGE_POISONING=y CONFIG_SLUB_DEBUG=y CONFIG_SLUB=y (the slub parts don't matter, ignore them) > > - _init_on_alloc_enabled_early = false; > > > - _init_on_free_enabled_early = false; > > > } > - if (_init_on_alloc_enabled_early) > - static_branch_enable(&init_on_alloc); > > > - else > - static_branch_disable(&init_on_alloc); > > > > - if (_init_on_free_enabled_early) > - static_branch_enable(&init_on_free); > > > - else > - static_branch_disable(&init_on_free); > > > > #ifdef CONFIG_DEBUG_PAGEALLOC > > if (!debug_pagealloc_enabled()) > > return; > > ------------------------------------------------------------------------ > > Kees Cook While both patches(Sergei's, and yours) work for me, I'll be using this one for now. Thank you for your work! Cheers! [-- Attachment #1.2: publickey - bowsingbetee@protonmail.com - 0x65EDD6AE.asc --] [-- Type: application/pgp-keys, Size: 1873 bytes --] [-- Attachment #2: OpenPGP digital signature --] [-- Type: application/pgp-signature, Size: 509 bytes --] ^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH] mm: page_alloc: fix page_poison=1 / INIT_ON_ALLOC_DEFAULT_ON interaction 2021-07-14 11:11 ` bowsingbetee @ 2021-07-14 21:14 ` Kees Cook 0 siblings, 0 replies; 7+ messages in thread From: Kees Cook @ 2021-07-14 21:14 UTC (permalink / raw) To: bowsingbetee Cc: Sergei Trofimovich, linux-mm, linux-kernel, Andrew Morton, Alexander Potapenko, Thomas Gleixner, Vlastimil Babka On Wed, Jul 14, 2021 at 11:11:18AM +0000, bowsingbetee wrote: > B & C variants show this message and A does not, which is what I would expect to happen. > > Tested variants: > > A. "page_poison=1 init_on_free=0 init_on_alloc=0 slub_debug=P" > B. "page_poison=1 init_on_free=0 init_on_alloc=0 slub_debug=P init_on_free=1" > C. "page_poison=1 slub_debug=P" > > in common: > CONFIG_INIT_ON_ALLOC_DEFAULT_ON=y > CONFIG_INIT_ON_FREE_DEFAULT_ON=y > CONFIG_PAGE_POISONING=y > CONFIG_SLUB_DEBUG=y > CONFIG_SLUB=y > > (the slub parts don't matter, ignore them) > > [...] > > While both patches(Sergei's, and yours) work for me, I'll be using this one for now. > > Thank you for your work! > Cheers! Awesome; thanks for testing it! -Kees -- Kees Cook ^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2021-07-17 18:20 UTC | newest] Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2021-07-11 23:57 v5.12+ regression on page_poison=1 i_on_free=0 i_on_alloc=0 warnings Sergei Trofimovich 2021-07-12 21:58 ` [PATCH] mm: page_alloc: fix page_poison=1 / INIT_ON_ALLOC_DEFAULT_ON interaction Sergei Trofimovich 2021-07-14 2:00 ` Andrew Morton 2021-07-17 18:18 ` Sergei Trofimovich 2021-07-14 3:10 ` Kees Cook 2021-07-14 11:11 ` bowsingbetee 2021-07-14 21:14 ` Kees Cook
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).