March 2023 - Linux-stable-mirror

[PATCH 5.10 2/3] kexec: turn all kexec_mutex acquisitions into trylocks

by wenyang.linux＠foxmail.com

From: Valentin Schneider <vschneid(a)redhat.com> commit 7bb5da0d490b2d836c5218f5186ee588d2145310 upstream. Patch series "kexec, panic: Making crash_kexec() NMI safe", v4. This patch (of 2): Most acquistions of kexec_mutex are done via mutex_trylock() - those were a direct "translation" from: 8c5a1cf0ad3a ("kexec: use a mutex for locking rather than xchg()") there have however been two additions since then that use mutex_lock(): crash_get_memory_size() and crash_shrink_memory(). A later commit will replace said mutex with an atomic variable, and locking operations will become atomic_cmpxchg(). Rather than having those mutex_lock() become while (atomic_cmpxchg(&lock, 0, 1)), turn them into trylocks that can return -EBUSY on acquisition failure. This does halve the printable size of the crash kernel, but that's still neighbouring 2G for 32bit kernels which should be ample enough. Link: https://lkml.kernel.org/r/20220630223258.4144112-1-vschneid@redhat.com Link: https://lkml.kernel.org/r/20220630223258.4144112-2-vschneid@redhat.com Signed-off-by: Valentin Schneider <vschneid(a)redhat.com> Cc: Arnd Bergmann <arnd(a)arndb.de> Cc: "Eric W . Biederman" <ebiederm(a)xmission.com> Cc: Juri Lelli <jlelli(a)redhat.com> Cc: Luis Claudio R. Goncalves <lgoncalv(a)redhat.com> Cc: Miaohe Lin <linmiaohe(a)huawei.com> Cc: Petr Mladek <pmladek(a)suse.com> Cc: Sebastian Andrzej Siewior <bigeasy(a)linutronix.de> Cc: Thomas Gleixner <tglx(a)linutronix.de> Cc: Baoquan He <bhe(a)redhat.com> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> Cc: stable(a)vger.kernel.org # 5.10+ Signed-off-by: Wen Yang <wenyang.linux(a)foxmail.com> --- include/linux/kexec.h | 2 +- kernel/kexec_core.c | 12 ++++++++---- kernel/ksysfs.c | 7 ++++++- 3 files changed, 15 insertions(+), 6 deletions(-) diff --git a/include/linux/kexec.h b/include/linux/kexec.h index a1f12e959bba..3c1deba496c9 100644 --- a/include/linux/kexec.h +++ b/include/linux/kexec.h @@ -380,8 +380,8 @@ extern note_buf_t __percpu *crash_notes; extern bool kexec_in_progress; int crash_shrink_memory(unsigned long new_size); -size_t crash_get_memory_size(void); void crash_free_reserved_phys_range(unsigned long begin, unsigned long end); +ssize_t crash_get_memory_size(void); void arch_kexec_protect_crashkres(void); void arch_kexec_unprotect_crashkres(void); diff --git a/kernel/kexec_core.c b/kernel/kexec_core.c index c589c7a9562c..e47870f30728 100644 --- a/kernel/kexec_core.c +++ b/kernel/kexec_core.c @@ -989,13 +989,16 @@ void crash_kexec(struct pt_regs *regs) } } -size_t crash_get_memory_size(void) +ssize_t crash_get_memory_size(void) { - size_t size = 0; + ssize_t size = 0; + + if (!mutex_trylock(&kexec_mutex)) + return -EBUSY; - mutex_lock(&kexec_mutex); if (crashk_res.end != crashk_res.start) size = resource_size(&crashk_res); + mutex_unlock(&kexec_mutex); return size; } @@ -1016,7 +1019,8 @@ int crash_shrink_memory(unsigned long new_size) unsigned long old_size; struct resource *ram_res; - mutex_lock(&kexec_mutex); + if (!mutex_trylock(&kexec_mutex)) + return -EBUSY; if (kexec_crash_image) { ret = -ENOENT; diff --git a/kernel/ksysfs.c b/kernel/ksysfs.c index 35859da8bd4f..e20c19e3ba49 100644 --- a/kernel/ksysfs.c +++ b/kernel/ksysfs.c @@ -106,7 +106,12 @@ KERNEL_ATTR_RO(kexec_crash_loaded); static ssize_t kexec_crash_size_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { - return sprintf(buf, "%zu\n", crash_get_memory_size()); + ssize_t size = crash_get_memory_size(); + + if (size < 0) + return size; + + return sprintf(buf, "%zd\n", size); } static ssize_t kexec_crash_size_store(struct kobject *kobj, struct kobj_attribute *attr, -- 2.37.2

1 year, 9 months

1
0
0 0

[PATCH 5.10 1/3] kexec: move locking into do_kexec_load

by wenyang.linux＠foxmail.com

From: Arnd Bergmann <arnd(a)arndb.de> commit 4b692e861619353ce069e547a67c8d0e32d9ef3d upstream. Patch series "compat: remove compat_alloc_user_space", v5. Going through compat_alloc_user_space() to convert indirect system call arguments tends to add complexity compared to handling the native and compat logic in the same code. This patch (of 6): The locking is the same between the native and compat version of sys_kexec_load(), so it can be done in the common implementation to reduce duplication. Link: https://lkml.kernel.org/r/20210727144859.4150043-1-arnd@kernel.org Link: https://lkml.kernel.org/r/20210727144859.4150043-2-arnd@kernel.org Signed-off-by: Arnd Bergmann <arnd(a)arndb.de> Co-developed-by: Eric Biederman <ebiederm(a)xmission.com> Co-developed-by: Christoph Hellwig <hch(a)infradead.org> Acked-by: "Eric W. Biederman" <ebiederm(a)xmission.com> Cc: Catalin Marinas <catalin.marinas(a)arm.com> Cc: Will Deacon <will(a)kernel.org> Cc: Thomas Bogendoerfer <tsbogend(a)alpha.franken.de> Cc: "James E.J. Bottomley" <James.Bottomley(a)HansenPartnership.com> Cc: Helge Deller <deller(a)gmx.de> Cc: Michael Ellerman <mpe(a)ellerman.id.au> Cc: Benjamin Herrenschmidt <benh(a)kernel.crashing.org> Cc: Paul Mackerras <paulus(a)samba.org> Cc: Heiko Carstens <hca(a)linux.ibm.com> Cc: Vasily Gorbik <gor(a)linux.ibm.com> Cc: Christian Borntraeger <borntraeger(a)de.ibm.com> Cc: "David S. Miller" <davem(a)davemloft.net> Cc: Thomas Gleixner <tglx(a)linutronix.de> Cc: Ingo Molnar <mingo(a)redhat.com> Cc: Borislav Petkov <bp(a)alien8.de> Cc: "H. Peter Anvin" <hpa(a)zytor.com> Cc: Al Viro <viro(a)zeniv.linux.org.uk> Cc: Feng Tang <feng.tang(a)intel.com> Cc: Christoph Hellwig <hch(a)lst.de> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds(a)linux-foundation.org> Cc: stable(a)vger.kernel.org # 5.10+ Signed-off-by: Wen Yang <wenyang.linux(a)foxmail.com> --- kernel/kexec.c | 44 ++++++++++++++++---------------------------- 1 file changed, 16 insertions(+), 28 deletions(-) diff --git a/kernel/kexec.c b/kernel/kexec.c index c82c6c06f051..9c7aef8f4bb6 100644 --- a/kernel/kexec.c +++ b/kernel/kexec.c @@ -110,6 +110,17 @@ static int do_kexec_load(unsigned long entry, unsigned long nr_segments, unsigned long i; int ret; + /* + * Because we write directly to the reserved memory region when loading + * crash kernels we need a mutex here to prevent multiple crash kernels + * from attempting to load simultaneously, and to prevent a crash kernel + * from loading over the top of a in use crash kernel. + * + * KISS: always take the mutex. + */ + if (!mutex_trylock(&kexec_mutex)) + return -EBUSY; + if (flags & KEXEC_ON_CRASH) { dest_image = &kexec_crash_image; if (kexec_crash_image) @@ -121,7 +132,8 @@ static int do_kexec_load(unsigned long entry, unsigned long nr_segments, if (nr_segments == 0) { /* Uninstall image */ kimage_free(xchg(dest_image, NULL)); - return 0; + ret = 0; + goto out_unlock; } if (flags & KEXEC_ON_CRASH) { /* @@ -134,7 +146,7 @@ static int do_kexec_load(unsigned long entry, unsigned long nr_segments, ret = kimage_alloc_init(&image, entry, nr_segments, segments, flags); if (ret) - return ret; + goto out_unlock; if (flags & KEXEC_PRESERVE_CONTEXT) image->preserve_context = 1; @@ -171,6 +183,8 @@ static int do_kexec_load(unsigned long entry, unsigned long nr_segments, arch_kexec_protect_crashkres(); kimage_free(image); +out_unlock: + mutex_unlock(&kexec_mutex); return ret; } @@ -247,21 +261,8 @@ SYSCALL_DEFINE4(kexec_load, unsigned long, entry, unsigned long, nr_segments, ((flags & KEXEC_ARCH_MASK) != KEXEC_ARCH_DEFAULT)) return -EINVAL; - /* Because we write directly to the reserved memory - * region when loading crash kernels we need a mutex here to - * prevent multiple crash kernels from attempting to load - * simultaneously, and to prevent a crash kernel from loading - * over the top of a in use crash kernel. - * - * KISS: always take the mutex. - */ - if (!mutex_trylock(&kexec_mutex)) - return -EBUSY; - result = do_kexec_load(entry, nr_segments, segments, flags); - mutex_unlock(&kexec_mutex); - return result; } @@ -301,21 +302,8 @@ COMPAT_SYSCALL_DEFINE4(kexec_load, compat_ulong_t, entry, return -EFAULT; } - /* Because we write directly to the reserved memory - * region when loading crash kernels we need a mutex here to - * prevent multiple crash kernels from attempting to load - * simultaneously, and to prevent a crash kernel from loading - * over the top of a in use crash kernel. - * - * KISS: always take the mutex. - */ - if (!mutex_trylock(&kexec_mutex)) - return -EBUSY; - result = do_kexec_load(entry, nr_segments, ksegments, flags); - mutex_unlock(&kexec_mutex); - return result; } #endif -- 2.37.2

1 year, 9 months

1
0
0 0

[PATCH 5.15 2/2] panic, kexec: make __crash_kexec() NMI safe

by wenyang.linux＠foxmail.com

From: Valentin Schneider <vschneid(a)redhat.com> commit 05c6257433b7212f07a7e53479a8ab038fc1666a upstream. Attempting to get a crash dump out of a debug PREEMPT_RT kernel via an NMI panic() doesn't work. The cause of that lies in the PREEMPT_RT definition of mutex_trylock(): if (IS_ENABLED(CONFIG_DEBUG_RT_MUTEXES) && WARN_ON_ONCE(!in_task())) return 0; This prevents an nmi_panic() from executing the main body of __crash_kexec() which does the actual kexec into the kdump kernel. The warning and return are explained by: 6ce47fd961fa ("rtmutex: Warn if trylock is called from hard/softirq context") [...] The reasons for this are: 1) There is a potential deadlock in the slowpath 2) Another cpu which blocks on the rtmutex will boost the task which allegedly locked the rtmutex, but that cannot work because the hard/softirq context borrows the task context. Furthermore, grabbing the lock isn't NMI safe, so do away with kexec_mutex and replace it with an atomic variable. This is somewhat overzealous as *some* callsites could keep using a mutex (e.g. the sysfs-facing ones like crash_shrink_memory()), but this has the benefit of involving a single unified lock and preventing any future NMI-related surprises. Tested by triggering NMI panics via: $ echo 1 > /proc/sys/kernel/panic_on_unrecovered_nmi $ echo 1 > /proc/sys/kernel/unknown_nmi_panic $ echo 1 > /proc/sys/kernel/panic $ ipmitool power diag Link: https://lkml.kernel.org/r/20220630223258.4144112-3-vschneid@redhat.com Fixes: 6ce47fd961fa ("rtmutex: Warn if trylock is called from hard/softirq context") Signed-off-by: Valentin Schneider <vschneid(a)redhat.com> Cc: Arnd Bergmann <arnd(a)arndb.de> Cc: Baoquan He <bhe(a)redhat.com> Cc: "Eric W . Biederman" <ebiederm(a)xmission.com> Cc: Juri Lelli <jlelli(a)redhat.com> Cc: Luis Claudio R. Goncalves <lgoncalv(a)redhat.com> Cc: Miaohe Lin <linmiaohe(a)huawei.com> Cc: Petr Mladek <pmladek(a)suse.com> Cc: Sebastian Andrzej Siewior <bigeasy(a)linutronix.de> Cc: Thomas Gleixner <tglx(a)linutronix.de> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> Cc: stable(a)vger.kernel.org # 5.15.x Signed-off-by: Wen Yang <wenyang.linux(a)foxmail.com> --- kernel/kexec.c | 11 ++++------- kernel/kexec_core.c | 20 ++++++++++---------- kernel/kexec_file.c | 4 ++-- kernel/kexec_internal.h | 15 ++++++++++++++- 4 files changed, 30 insertions(+), 20 deletions(-) diff --git a/kernel/kexec.c b/kernel/kexec.c index b5e40f069768..cb8e6e6f983c 100644 --- a/kernel/kexec.c +++ b/kernel/kexec.c @@ -93,13 +93,10 @@ static int do_kexec_load(unsigned long entry, unsigned long nr_segments, /* * Because we write directly to the reserved memory region when loading - * crash kernels we need a mutex here to prevent multiple crash kernels - * from attempting to load simultaneously, and to prevent a crash kernel - * from loading over the top of a in use crash kernel. - * - * KISS: always take the mutex. + * crash kernels we need a serialization here to prevent multiple crash + * kernels from attempting to load simultaneously. */ - if (!mutex_trylock(&kexec_mutex)) + if (!kexec_trylock()) return -EBUSY; if (flags & KEXEC_ON_CRASH) { @@ -165,7 +162,7 @@ static int do_kexec_load(unsigned long entry, unsigned long nr_segments, kimage_free(image); out_unlock: - mutex_unlock(&kexec_mutex); + kexec_unlock(); return ret; } diff --git a/kernel/kexec_core.c b/kernel/kexec_core.c index a101d2b77936..bdc2d952911c 100644 --- a/kernel/kexec_core.c +++ b/kernel/kexec_core.c @@ -46,7 +46,7 @@ #include <crypto/hash.h> #include "kexec_internal.h" -DEFINE_MUTEX(kexec_mutex); +atomic_t __kexec_lock = ATOMIC_INIT(0); /* Per cpu memory for storing cpu states in case of system crash. */ note_buf_t __percpu *crash_notes; @@ -944,7 +944,7 @@ int kexec_load_disabled; */ void __noclone __crash_kexec(struct pt_regs *regs) { - /* Take the kexec_mutex here to prevent sys_kexec_load + /* Take the kexec_lock here to prevent sys_kexec_load * running on one cpu from replacing the crash kernel * we are using after a panic on a different cpu. * @@ -952,7 +952,7 @@ void __noclone __crash_kexec(struct pt_regs *regs) * of memory the xchg(&kexec_crash_image) would be * sufficient. But since I reuse the memory... */ - if (mutex_trylock(&kexec_mutex)) { + if (kexec_trylock()) { if (kexec_crash_image) { struct pt_regs fixed_regs; @@ -961,7 +961,7 @@ void __noclone __crash_kexec(struct pt_regs *regs) machine_crash_shutdown(&fixed_regs); machine_kexec(kexec_crash_image); } - mutex_unlock(&kexec_mutex); + kexec_unlock(); } } STACK_FRAME_NON_STANDARD(__crash_kexec); @@ -993,13 +993,13 @@ ssize_t crash_get_memory_size(void) { ssize_t size = 0; - if (!mutex_trylock(&kexec_mutex)) + if (!kexec_trylock()) return -EBUSY; if (crashk_res.end != crashk_res.start) size = resource_size(&crashk_res); - mutex_unlock(&kexec_mutex); + kexec_unlock(); return size; } @@ -1019,7 +1019,7 @@ int crash_shrink_memory(unsigned long new_size) unsigned long old_size; struct resource *ram_res; - if (!mutex_trylock(&kexec_mutex)) + if (!kexec_trylock()) return -EBUSY; if (kexec_crash_image) { @@ -1058,7 +1058,7 @@ int crash_shrink_memory(unsigned long new_size) insert_resource(&iomem_resource, ram_res); unlock: - mutex_unlock(&kexec_mutex); + kexec_unlock(); return ret; } @@ -1130,7 +1130,7 @@ int kernel_kexec(void) { int error = 0; - if (!mutex_trylock(&kexec_mutex)) + if (!kexec_trylock()) return -EBUSY; if (!kexec_image) { error = -EINVAL; @@ -1206,7 +1206,7 @@ int kernel_kexec(void) #endif Unlock: - mutex_unlock(&kexec_mutex); + kexec_unlock(); return error; } diff --git a/kernel/kexec_file.c b/kernel/kexec_file.c index f7a4fd4d243f..1fb7ff690577 100644 --- a/kernel/kexec_file.c +++ b/kernel/kexec_file.c @@ -343,7 +343,7 @@ SYSCALL_DEFINE5(kexec_file_load, int, kernel_fd, int, initrd_fd, image = NULL; - if (!mutex_trylock(&kexec_mutex)) + if (!kexec_trylock()) return -EBUSY; dest_image = &kexec_image; @@ -415,7 +415,7 @@ SYSCALL_DEFINE5(kexec_file_load, int, kernel_fd, int, initrd_fd, if ((flags & KEXEC_FILE_ON_CRASH) && kexec_crash_image) arch_kexec_protect_crashkres(); - mutex_unlock(&kexec_mutex); + kexec_unlock(); kimage_free(image); return ret; } diff --git a/kernel/kexec_internal.h b/kernel/kexec_internal.h index 48aaf2ac0d0d..74da1409cd14 100644 --- a/kernel/kexec_internal.h +++ b/kernel/kexec_internal.h @@ -13,7 +13,20 @@ void kimage_terminate(struct kimage *image); int kimage_is_destination_range(struct kimage *image, unsigned long start, unsigned long end); -extern struct mutex kexec_mutex; +/* + * Whatever is used to serialize accesses to the kexec_crash_image needs to be + * NMI safe, as __crash_kexec() can happen during nmi_panic(), so here we use a + * "simple" atomic variable that is acquired with a cmpxchg(). + */ +extern atomic_t __kexec_lock; +static inline bool kexec_trylock(void) +{ + return atomic_cmpxchg_acquire(&__kexec_lock, 0, 1) == 0; +} +static inline void kexec_unlock(void) +{ + atomic_set_release(&__kexec_lock, 0); +} #ifdef CONFIG_KEXEC_FILE #include <linux/purgatory.h> -- 2.37.2

1 year, 9 months

1
0
0 0

[PATCH 5.15 1/2] kexec: turn all kexec_mutex acquisitions into trylocks

by wenyang.linux＠foxmail.com

From: Valentin Schneider <vschneid(a)redhat.com> commit 7bb5da0d490b2d836c5218f5186ee588d2145310 upstream. Patch series "kexec, panic: Making crash_kexec() NMI safe", v4. This patch (of 2): Most acquistions of kexec_mutex are done via mutex_trylock() - those were a direct "translation" from: 8c5a1cf0ad3a ("kexec: use a mutex for locking rather than xchg()") there have however been two additions since then that use mutex_lock(): crash_get_memory_size() and crash_shrink_memory(). A later commit will replace said mutex with an atomic variable, and locking operations will become atomic_cmpxchg(). Rather than having those mutex_lock() become while (atomic_cmpxchg(&lock, 0, 1)), turn them into trylocks that can return -EBUSY on acquisition failure. This does halve the printable size of the crash kernel, but that's still neighbouring 2G for 32bit kernels which should be ample enough. Link: https://lkml.kernel.org/r/20220630223258.4144112-1-vschneid@redhat.com Link: https://lkml.kernel.org/r/20220630223258.4144112-2-vschneid@redhat.com Signed-off-by: Valentin Schneider <vschneid(a)redhat.com> Cc: Arnd Bergmann <arnd(a)arndb.de> Cc: "Eric W . Biederman" <ebiederm(a)xmission.com> Cc: Juri Lelli <jlelli(a)redhat.com> Cc: Luis Claudio R. Goncalves <lgoncalv(a)redhat.com> Cc: Miaohe Lin <linmiaohe(a)huawei.com> Cc: Petr Mladek <pmladek(a)suse.com> Cc: Sebastian Andrzej Siewior <bigeasy(a)linutronix.de> Cc: Thomas Gleixner <tglx(a)linutronix.de> Cc: Baoquan He <bhe(a)redhat.com> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> Cc: <stable(a)vger.kernel.org> # v5.15+ Signed-off-by: Wen Yang <wenyang.linux(a)foxmail.com> --- include/linux/kexec.h | 2 +- kernel/kexec_core.c | 12 ++++++++---- kernel/ksysfs.c | 7 ++++++- 3 files changed, 15 insertions(+), 6 deletions(-) diff --git a/include/linux/kexec.h b/include/linux/kexec.h index cf042d41c87b..88c289ce3039 100644 --- a/include/linux/kexec.h +++ b/include/linux/kexec.h @@ -390,8 +390,8 @@ extern note_buf_t __percpu *crash_notes; extern bool kexec_in_progress; int crash_shrink_memory(unsigned long new_size); -size_t crash_get_memory_size(void); void crash_free_reserved_phys_range(unsigned long begin, unsigned long end); +ssize_t crash_get_memory_size(void); void arch_kexec_protect_crashkres(void); void arch_kexec_unprotect_crashkres(void); diff --git a/kernel/kexec_core.c b/kernel/kexec_core.c index 5a5d192a89ac..a101d2b77936 100644 --- a/kernel/kexec_core.c +++ b/kernel/kexec_core.c @@ -989,13 +989,16 @@ void crash_kexec(struct pt_regs *regs) } } -size_t crash_get_memory_size(void) +ssize_t crash_get_memory_size(void) { - size_t size = 0; + ssize_t size = 0; + + if (!mutex_trylock(&kexec_mutex)) + return -EBUSY; - mutex_lock(&kexec_mutex); if (crashk_res.end != crashk_res.start) size = resource_size(&crashk_res); + mutex_unlock(&kexec_mutex); return size; } @@ -1016,7 +1019,8 @@ int crash_shrink_memory(unsigned long new_size) unsigned long old_size; struct resource *ram_res; - mutex_lock(&kexec_mutex); + if (!mutex_trylock(&kexec_mutex)) + return -EBUSY; if (kexec_crash_image) { ret = -ENOENT; diff --git a/kernel/ksysfs.c b/kernel/ksysfs.c index 35859da8bd4f..e20c19e3ba49 100644 --- a/kernel/ksysfs.c +++ b/kernel/ksysfs.c @@ -106,7 +106,12 @@ KERNEL_ATTR_RO(kexec_crash_loaded); static ssize_t kexec_crash_size_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { - return sprintf(buf, "%zu\n", crash_get_memory_size()); + ssize_t size = crash_get_memory_size(); + + if (size < 0) + return size; + + return sprintf(buf, "%zd\n", size); } static ssize_t kexec_crash_size_store(struct kobject *kobj, struct kobj_attribute *attr, -- 2.37.2

1 year, 9 months

1
0
0 0

[PATCH 3/3] panic, kexec: make __crash_kexec() NMI safe

by wenyang.linux＠foxmail.com

From: Valentin Schneider <vschneid(a)redhat.com> commit 811d581194f7412eda97acc03d17fc77824b561f upstream. Attempting to get a crash dump out of a debug PREEMPT_RT kernel via an NMI panic() doesn't work. The cause of that lies in the PREEMPT_RT definition of mutex_trylock(): if (IS_ENABLED(CONFIG_DEBUG_RT_MUTEXES) && WARN_ON_ONCE(!in_task())) return 0; This prevents an nmi_panic() from executing the main body of __crash_kexec() which does the actual kexec into the kdump kernel. The warning and return are explained by: 6ce47fd961fa ("rtmutex: Warn if trylock is called from hard/softirq context") [...] The reasons for this are: 1) There is a potential deadlock in the slowpath 2) Another cpu which blocks on the rtmutex will boost the task which allegedly locked the rtmutex, but that cannot work because the hard/softirq context borrows the task context. Furthermore, grabbing the lock isn't NMI safe, so do away with kexec_mutex and replace it with an atomic variable. This is somewhat overzealous as *some* callsites could keep using a mutex (e.g. the sysfs-facing ones like crash_shrink_memory()), but this has the benefit of involving a single unified lock and preventing any future NMI-related surprises. Tested by triggering NMI panics via: $ echo 1 > /proc/sys/kernel/panic_on_unrecovered_nmi $ echo 1 > /proc/sys/kernel/unknown_nmi_panic $ echo 1 > /proc/sys/kernel/panic $ ipmitool power diag Link: https://lkml.kernel.org/r/20220630223258.4144112-3-vschneid@redhat.com Fixes: 6ce47fd961fa ("rtmutex: Warn if trylock is called from hard/softirq context") Signed-off-by: Valentin Schneider <vschneid(a)redhat.com> Cc: Arnd Bergmann <arnd(a)arndb.de> Cc: Baoquan He <bhe(a)redhat.com> Cc: "Eric W . Biederman" <ebiederm(a)xmission.com> Cc: Juri Lelli <jlelli(a)redhat.com> Cc: Luis Claudio R. Goncalves <lgoncalv(a)redhat.com> Cc: Miaohe Lin <linmiaohe(a)huawei.com> Cc: Petr Mladek <pmladek(a)suse.com> Cc: Sebastian Andrzej Siewior <bigeasy(a)linutronix.de> Cc: Thomas Gleixner <tglx(a)linutronix.de> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> Cc: stable(a)vger.kernel.org # 5.10+ Signed-off-by: Wen Yang <wenyang.linux(a)foxmail.com> --- kernel/kexec.c | 11 ++++------- kernel/kexec_core.c | 20 ++++++++++---------- kernel/kexec_file.c | 4 ++-- kernel/kexec_internal.h | 15 ++++++++++++++- 4 files changed, 30 insertions(+), 20 deletions(-) diff --git a/kernel/kexec.c b/kernel/kexec.c index 9c7aef8f4bb6..f0f0c6555454 100644 --- a/kernel/kexec.c +++ b/kernel/kexec.c @@ -112,13 +112,10 @@ static int do_kexec_load(unsigned long entry, unsigned long nr_segments, /* * Because we write directly to the reserved memory region when loading - * crash kernels we need a mutex here to prevent multiple crash kernels - * from attempting to load simultaneously, and to prevent a crash kernel - * from loading over the top of a in use crash kernel. - * - * KISS: always take the mutex. + * crash kernels we need a serialization here to prevent multiple crash + * kernels from attempting to load simultaneously. */ - if (!mutex_trylock(&kexec_mutex)) + if (!kexec_trylock()) return -EBUSY; if (flags & KEXEC_ON_CRASH) { @@ -184,7 +181,7 @@ static int do_kexec_load(unsigned long entry, unsigned long nr_segments, kimage_free(image); out_unlock: - mutex_unlock(&kexec_mutex); + kexec_unlock(); return ret; } diff --git a/kernel/kexec_core.c b/kernel/kexec_core.c index e47870f30728..7a8104d48997 100644 --- a/kernel/kexec_core.c +++ b/kernel/kexec_core.c @@ -45,7 +45,7 @@ #include <crypto/sha.h> #include "kexec_internal.h" -DEFINE_MUTEX(kexec_mutex); +atomic_t __kexec_lock = ATOMIC_INIT(0); /* Per cpu memory for storing cpu states in case of system crash. */ note_buf_t __percpu *crash_notes; @@ -943,7 +943,7 @@ int kexec_load_disabled; */ void __noclone __crash_kexec(struct pt_regs *regs) { - /* Take the kexec_mutex here to prevent sys_kexec_load + /* Take the kexec_lock here to prevent sys_kexec_load * running on one cpu from replacing the crash kernel * we are using after a panic on a different cpu. * @@ -951,7 +951,7 @@ void __noclone __crash_kexec(struct pt_regs *regs) * of memory the xchg(&kexec_crash_image) would be * sufficient. But since I reuse the memory... */ - if (mutex_trylock(&kexec_mutex)) { + if (kexec_trylock()) { if (kexec_crash_image) { struct pt_regs fixed_regs; @@ -960,7 +960,7 @@ void __noclone __crash_kexec(struct pt_regs *regs) machine_crash_shutdown(&fixed_regs); machine_kexec(kexec_crash_image); } - mutex_unlock(&kexec_mutex); + kexec_unlock(); } } STACK_FRAME_NON_STANDARD(__crash_kexec); @@ -993,13 +993,13 @@ ssize_t crash_get_memory_size(void) { ssize_t size = 0; - if (!mutex_trylock(&kexec_mutex)) + if (!kexec_trylock()) return -EBUSY; if (crashk_res.end != crashk_res.start) size = resource_size(&crashk_res); - mutex_unlock(&kexec_mutex); + kexec_unlock(); return size; } @@ -1019,7 +1019,7 @@ int crash_shrink_memory(unsigned long new_size) unsigned long old_size; struct resource *ram_res; - if (!mutex_trylock(&kexec_mutex)) + if (!kexec_trylock()) return -EBUSY; if (kexec_crash_image) { @@ -1058,7 +1058,7 @@ int crash_shrink_memory(unsigned long new_size) insert_resource(&iomem_resource, ram_res); unlock: - mutex_unlock(&kexec_mutex); + kexec_unlock(); return ret; } @@ -1130,7 +1130,7 @@ int kernel_kexec(void) { int error = 0; - if (!mutex_trylock(&kexec_mutex)) + if (!kexec_trylock()) return -EBUSY; if (!kexec_image) { error = -EINVAL; @@ -1205,7 +1205,7 @@ int kernel_kexec(void) #endif Unlock: - mutex_unlock(&kexec_mutex); + kexec_unlock(); return error; } diff --git a/kernel/kexec_file.c b/kernel/kexec_file.c index fff11916aba3..b9c857782ada 100644 --- a/kernel/kexec_file.c +++ b/kernel/kexec_file.c @@ -343,7 +343,7 @@ SYSCALL_DEFINE5(kexec_file_load, int, kernel_fd, int, initrd_fd, image = NULL; - if (!mutex_trylock(&kexec_mutex)) + if (!kexec_trylock()) return -EBUSY; dest_image = &kexec_image; @@ -415,7 +415,7 @@ SYSCALL_DEFINE5(kexec_file_load, int, kernel_fd, int, initrd_fd, if ((flags & KEXEC_FILE_ON_CRASH) && kexec_crash_image) arch_kexec_protect_crashkres(); - mutex_unlock(&kexec_mutex); + kexec_unlock(); kimage_free(image); return ret; } diff --git a/kernel/kexec_internal.h b/kernel/kexec_internal.h index 39d30ccf8d87..49d4e3ab9c96 100644 --- a/kernel/kexec_internal.h +++ b/kernel/kexec_internal.h @@ -15,7 +15,20 @@ int kimage_is_destination_range(struct kimage *image, int machine_kexec_post_load(struct kimage *image); -extern struct mutex kexec_mutex; +/* + * Whatever is used to serialize accesses to the kexec_crash_image needs to be + * NMI safe, as __crash_kexec() can happen during nmi_panic(), so here we use a + * "simple" atomic variable that is acquired with a cmpxchg(). + */ +extern atomic_t __kexec_lock; +static inline bool kexec_trylock(void) +{ + return atomic_cmpxchg_acquire(&__kexec_lock, 0, 1) == 0; +} +static inline void kexec_unlock(void) +{ + atomic_set_release(&__kexec_lock, 0); +} #ifdef CONFIG_KEXEC_FILE #include <linux/purgatory.h> -- 2.37.2

1 year, 9 months

1
0
0 0

[PATCH 2/3] kexec: turn all kexec_mutex acquisitions into trylocks

by wenyang.linux＠foxmail.com

From: Valentin Schneider <vschneid(a)redhat.com> commit 7bb5da0d490b2d836c5218f5186ee588d2145310 upstream. Patch series "kexec, panic: Making crash_kexec() NMI safe", v4. This patch (of 2): Most acquistions of kexec_mutex are done via mutex_trylock() - those were a direct "translation" from: 8c5a1cf0ad3a ("kexec: use a mutex for locking rather than xchg()") there have however been two additions since then that use mutex_lock(): crash_get_memory_size() and crash_shrink_memory(). A later commit will replace said mutex with an atomic variable, and locking operations will become atomic_cmpxchg(). Rather than having those mutex_lock() become while (atomic_cmpxchg(&lock, 0, 1)), turn them into trylocks that can return -EBUSY on acquisition failure. This does halve the printable size of the crash kernel, but that's still neighbouring 2G for 32bit kernels which should be ample enough. Link: https://lkml.kernel.org/r/20220630223258.4144112-1-vschneid@redhat.com Link: https://lkml.kernel.org/r/20220630223258.4144112-2-vschneid@redhat.com Signed-off-by: Valentin Schneider <vschneid(a)redhat.com> Cc: Arnd Bergmann <arnd(a)arndb.de> Cc: "Eric W . Biederman" <ebiederm(a)xmission.com> Cc: Juri Lelli <jlelli(a)redhat.com> Cc: Luis Claudio R. Goncalves <lgoncalv(a)redhat.com> Cc: Miaohe Lin <linmiaohe(a)huawei.com> Cc: Petr Mladek <pmladek(a)suse.com> Cc: Sebastian Andrzej Siewior <bigeasy(a)linutronix.de> Cc: Thomas Gleixner <tglx(a)linutronix.de> Cc: Baoquan He <bhe(a)redhat.com> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> Cc: stable(a)vger.kernel.org # 5.10+ Signed-off-by: Wen Yang <wenyang.linux(a)foxmail.com> --- include/linux/kexec.h | 2 +- kernel/kexec_core.c | 12 ++++++++---- kernel/ksysfs.c | 7 ++++++- 3 files changed, 15 insertions(+), 6 deletions(-) diff --git a/include/linux/kexec.h b/include/linux/kexec.h index a1f12e959bba..3c1deba496c9 100644 --- a/include/linux/kexec.h +++ b/include/linux/kexec.h @@ -380,8 +380,8 @@ extern note_buf_t __percpu *crash_notes; extern bool kexec_in_progress; int crash_shrink_memory(unsigned long new_size); -size_t crash_get_memory_size(void); void crash_free_reserved_phys_range(unsigned long begin, unsigned long end); +ssize_t crash_get_memory_size(void); void arch_kexec_protect_crashkres(void); void arch_kexec_unprotect_crashkres(void); diff --git a/kernel/kexec_core.c b/kernel/kexec_core.c index c589c7a9562c..e47870f30728 100644 --- a/kernel/kexec_core.c +++ b/kernel/kexec_core.c @@ -989,13 +989,16 @@ void crash_kexec(struct pt_regs *regs) } } -size_t crash_get_memory_size(void) +ssize_t crash_get_memory_size(void) { - size_t size = 0; + ssize_t size = 0; + + if (!mutex_trylock(&kexec_mutex)) + return -EBUSY; - mutex_lock(&kexec_mutex); if (crashk_res.end != crashk_res.start) size = resource_size(&crashk_res); + mutex_unlock(&kexec_mutex); return size; } @@ -1016,7 +1019,8 @@ int crash_shrink_memory(unsigned long new_size) unsigned long old_size; struct resource *ram_res; - mutex_lock(&kexec_mutex); + if (!mutex_trylock(&kexec_mutex)) + return -EBUSY; if (kexec_crash_image) { ret = -ENOENT; diff --git a/kernel/ksysfs.c b/kernel/ksysfs.c index 35859da8bd4f..e20c19e3ba49 100644 --- a/kernel/ksysfs.c +++ b/kernel/ksysfs.c @@ -106,7 +106,12 @@ KERNEL_ATTR_RO(kexec_crash_loaded); static ssize_t kexec_crash_size_show(struct kobject *kobj, struct kobj_attribute *attr, char *buf) { - return sprintf(buf, "%zu\n", crash_get_memory_size()); + ssize_t size = crash_get_memory_size(); + + if (size < 0) + return size; + + return sprintf(buf, "%zd\n", size); } static ssize_t kexec_crash_size_store(struct kobject *kobj, struct kobj_attribute *attr, -- 2.37.2

1 year, 9 months

1
0
0 0

[PATCH stable-queue.git] scripts/bad_stable: include instructions to fix and resubmit

by Vegard Nossum

This basically implements the suggestion from <https://lore.kernel.org/stable/46577540-6ed7-0ff5-11d5-2982cd29e92b@oracle.…>. Sample output: To reproduce the conflict and resubmit, you may use the following commands: git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-5.15.y git checkout FETCH_HEAD git cherry-pick -x 9f46c187e2e680ecd9de7983e4d081c3391acc76 # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '<165314153515625(a)kroah.com>' --subject-prefix 'PATCH 5.15.y' HEAD^.. Signed-off-by: Vegard Nossum <vegard.nossum(a)oracle.com> --- scripts/bad_stable | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/scripts/bad_stable b/scripts/bad_stable index 03165f8212..0cd0319fb3 100755 --- a/scripts/bad_stable +++ b/scripts/bad_stable @@ -136,6 +136,15 @@ reply() echo "tree, then please email the backport, including the original git commit" echo "id to <stable(a)vger.kernel.org>." echo + echo "To reproduce the conflict and resubmit, you may use the following commands:" + echo + echo "git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-$STABLE_VERSION.y" + echo "git checkout FETCH_HEAD" + echo "git cherry-pick -x $git_id" + echo "# <resolve conflicts, build, test, etc.>" + echo "git commit -s" + echo "git send-email --to '$STABLE' --in-reply-to '$ID' --subject-prefix 'PATCH $STABLE_VERSION.y' HEAD^.." + echo echo "Possible dependencies:" echo echo "$(curl --fail --silent https://git.kernel.org/pub/scm/linux/kernel/git/sashal/deps.git/plain/v$STA…)" -- 2.35.1.46.g38062e73e0

1 year, 9 months

2
1
0 0

[PATCH 5.10 1/2] drm/virtio: Fix NULL vs IS_ERR checking in virtio_gpu_object_shmem_init

by Ovidiu Panait

From: Miaoqian Lin <linmq006(a)gmail.com> commit c24968734abfed81c8f93dc5f44a7b7a9aecadfa upstream. Since drm_prime_pages_to_sg() function return error pointers. The drm_gem_shmem_get_sg_table() function returns error pointers too. Using IS_ERR() to check the return value to fix this. Fixes: 2f2aa13724d5 ("drm/virtio: move virtio_gpu_mem_entry initialization to new function") Signed-off-by: Miaoqian Lin <linmq006(a)gmail.com> Link: http://patchwork.freedesktop.org/patch/msgid/20220602104223.54527-1-linmq00… Signed-off-by: Gerd Hoffmann <kraxel(a)redhat.com> Signed-off-by: Ovidiu Panait <ovidiu.panait(a)windriver.com> --- drivers/gpu/drm/virtio/virtgpu_object.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/virtio/virtgpu_object.c b/drivers/gpu/drm/virtio/virtgpu_object.c index 0c98978e2e55..d4fab3361d2c 100644 --- a/drivers/gpu/drm/virtio/virtgpu_object.c +++ b/drivers/gpu/drm/virtio/virtgpu_object.c @@ -157,9 +157,9 @@ static int virtio_gpu_object_shmem_init(struct virtio_gpu_device *vgdev, * since virtio_gpu doesn't support dma-buf import from other devices. */ shmem->pages = drm_gem_shmem_get_sg_table(&bo->base.base); - if (!shmem->pages) { + if (IS_ERR(shmem->pages)) { drm_gem_shmem_unpin(&bo->base.base); - return -EINVAL; + return PTR_ERR(shmem->pages); } if (use_dma_api) { -- 2.39.1

1 year, 9 months

2
2
0 0

[PATCH for 4.14,4.19,5.4] dmaengine: sh: rcar-dmac: Check for error num after dma_set_max_seg_size

by Nobuhiro Iwamatsu

From: Jiasheng Jiang <jiasheng(a)iscas.ac.cn> commit da2ad87fba0891576aadda9161b8505fde81a84d upstream. As the possible failure of the dma_set_max_seg_size(), it should be better to check the return value of the dma_set_max_seg_size(). Fixes: 97d49c59e219 ("dmaengine: rcar-dmac: set scatter/gather max segment size") Reported-by: Geert Uytterhoeven <geert+renesas(a)glider.be> Signed-off-by: Jiasheng Jiang <jiasheng(a)iscas.ac.cn> Reviewed-by: Geert Uytterhoeven <geert+renesas(a)glider.be> Link: https://lore.kernel.org/r/20220111011239.452837-1-jiasheng@iscas.ac.cn Signed-off-by: Vinod Koul <vkoul(a)kernel.org> Signed-off-by: Nobuhiro Iwamatsu (CIP) <nobuhiro1.iwamatsu(a)toshiba.co.jp> --- drivers/dma/sh/rcar-dmac.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/drivers/dma/sh/rcar-dmac.c b/drivers/dma/sh/rcar-dmac.c index eba942441e3823..10a8a6d4e86015 100644 --- a/drivers/dma/sh/rcar-dmac.c +++ b/drivers/dma/sh/rcar-dmac.c @@ -1824,7 +1824,10 @@ static int rcar_dmac_probe(struct platform_device *pdev) dmac->dev = &pdev->dev; platform_set_drvdata(pdev, dmac); dmac->dev->dma_parms = &dmac->parms; - dma_set_max_seg_size(dmac->dev, RCAR_DMATCR_MASK); + ret = dma_set_max_seg_size(dmac->dev, RCAR_DMATCR_MASK); + if (ret) + return ret; + ret = dma_set_mask_and_coherent(dmac->dev, DMA_BIT_MASK(40)); if (ret) return ret; -- 2.36.1

1 year, 9 months

2
1
0 0

FAILED: patch "[PATCH] scripts/tags.sh: fix incompatibility with PCRE2" failed to apply to 5.10-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 5.10-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. Possible dependencies: 6ec363fc6142 ("scripts/tags.sh: fix incompatibility with PCRE2") 7394d2ebb651 ("scripts/tags.sh: Invoke 'realpath' via 'xargs'") thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 6ec363fc6142226b9ab5a6528f65333d729d2b6b Mon Sep 17 00:00:00 2001 From: Carlos Llamas <cmllamas(a)google.com> Date: Wed, 15 Feb 2023 18:38:50 +0000 Subject: [PATCH] scripts/tags.sh: fix incompatibility with PCRE2 Starting with release 10.38 PCRE2 drops default support for using \K in lookaround patterns as described in [1]. Unfortunately, scripts/tags.sh relies on such functionality to collect all_compiled_soures() leading to the following error: $ make COMPILED_SOURCE=1 tags GEN tags grep: \K is not allowed in lookarounds (but see PCRE2_EXTRA_ALLOW_LOOKAROUND_BSK) The usage of \K for this pattern was introduced in commit 4f491bb6ea2a ("scripts/tags.sh: collect compiled source precisely") which speeds up the generation of tags significantly. In order to fix this issue without compromising the performance we can switch over to an equivalent sed expression. The same matching pattern is preserved here except \K is replaced with a backreference \1. [1] https://www.pcre.org/current/doc/html/pcre2syntax.html#SEC11 Cc: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> Cc: Cristian Ciocaltea <cristian.ciocaltea(a)collabora.com> Cc: Masahiro Yamada <masahiroy(a)kernel.org> Cc: Jialu Xu <xujialu(a)vimux.org> Cc: Vipin Sharma <vipinsh(a)google.com> Cc: stable(a)vger.kernel.org Fixes: 4f491bb6ea2a ("scripts/tags.sh: collect compiled source precisely") Signed-off-by: Carlos Llamas <cmllamas(a)google.com> Link: https://lore.kernel.org/r/20230215183850.3353198-1-cmllamas@google.com Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> diff --git a/scripts/tags.sh b/scripts/tags.sh index 1ad45f17179a..6b9001853890 100755 --- a/scripts/tags.sh +++ b/scripts/tags.sh @@ -98,7 +98,7 @@ all_compiled_sources() { echo include/generated/autoconf.h find $ignore -name "*.cmd" -exec \ - grep -Poh '(?(?=^source_.* \K).*|(?=^ \K\S).*(?= \\))' {} \+ | + sed -n -E 's/^source_.* (.*)/\1/p; s/^ (\S.*) \\/\1/p' {} \+ | awk '!a[$0]++' } | xargs realpath -esq $([ -z "$KBUILD_ABS_SRCTREE" ] && echo --relative-to=.) | sort -u

1 year, 9 months

3
2
0 0

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror March 2023