From 161485e8273001e56b2f20755ad9b6217b601fb3 Mon Sep 17 00:00:00 2001 From: Ard Biesheuvel Date: Mon, 4 Aug 2014 18:16:00 +0200 Subject: efi: Implement mandatory locking for UEFI Runtime Services According to section 7.1 of the UEFI spec, Runtime Services are not fully reentrant, and there are particular combinations of calls that need to be serialized. Use a spinlock to serialize all Runtime Services with respect to all others, even if this is more than strictly needed. We've managed to get away without requiring a runtime services lock until now because most of the interactions with EFI involve EFI variables, and those operations are already serialised with __efivars->lock. Some of the assumptions underlying the decision whether locks are needed or not (e.g., SetVariable() against ResetSystem()) may not apply universally to all [new] architectures that implement UEFI. Rather than try to reason our way out of this, let's just implement at least what the spec requires in terms of locking. Signed-off-by: Ard Biesheuvel Signed-off-by: Matt Fleming --- drivers/firmware/efi/runtime-wrappers.c | 154 +++++++++++++++++++++++++++++--- 1 file changed, 144 insertions(+), 10 deletions(-) (limited to 'drivers/firmware/efi/runtime-wrappers.c') diff --git a/drivers/firmware/efi/runtime-wrappers.c b/drivers/firmware/efi/runtime-wrappers.c index 10daa4bbb258..9694cba665c4 100644 --- a/drivers/firmware/efi/runtime-wrappers.c +++ b/drivers/firmware/efi/runtime-wrappers.c @@ -14,10 +14,82 @@ * This file is released under the GPLv2. */ +#include #include -#include /* spinlock_t */ +#include +#include #include +/* + * According to section 7.1 of the UEFI spec, Runtime Services are not fully + * reentrant, and there are particular combinations of calls that need to be + * serialized. (source: UEFI Specification v2.4A) + * + * Table 31. Rules for Reentry Into Runtime Services + * +------------------------------------+-------------------------------+ + * | If previous call is busy in | Forbidden to call | + * +------------------------------------+-------------------------------+ + * | Any | SetVirtualAddressMap() | + * +------------------------------------+-------------------------------+ + * | ConvertPointer() | ConvertPointer() | + * +------------------------------------+-------------------------------+ + * | SetVariable() | ResetSystem() | + * | UpdateCapsule() | | + * | SetTime() | | + * | SetWakeupTime() | | + * | GetNextHighMonotonicCount() | | + * +------------------------------------+-------------------------------+ + * | GetVariable() | GetVariable() | + * | GetNextVariableName() | GetNextVariableName() | + * | SetVariable() | SetVariable() | + * | QueryVariableInfo() | QueryVariableInfo() | + * | UpdateCapsule() | UpdateCapsule() | + * | QueryCapsuleCapabilities() | QueryCapsuleCapabilities() | + * | GetNextHighMonotonicCount() | GetNextHighMonotonicCount() | + * +------------------------------------+-------------------------------+ + * | GetTime() | GetTime() | + * | SetTime() | SetTime() | + * | GetWakeupTime() | GetWakeupTime() | + * | SetWakeupTime() | SetWakeupTime() | + * +------------------------------------+-------------------------------+ + * + * Due to the fact that the EFI pstore may write to the variable store in + * interrupt context, we need to use a spinlock for at least the groups that + * contain SetVariable() and QueryVariableInfo(). That leaves little else, as + * none of the remaining functions are actually ever called at runtime. + * So let's just use a single spinlock to serialize all Runtime Services calls. + */ +static DEFINE_SPINLOCK(efi_runtime_lock); + +/* + * Some runtime services calls can be reentrant under NMI, even if the table + * above says they are not. (source: UEFI Specification v2.4A) + * + * Table 32. Functions that may be called after Machine Check, INIT and NMI + * +----------------------------+------------------------------------------+ + * | Function | Called after Machine Check, INIT and NMI | + * +----------------------------+------------------------------------------+ + * | GetTime() | Yes, even if previously busy. | + * | GetVariable() | Yes, even if previously busy | + * | GetNextVariableName() | Yes, even if previously busy | + * | QueryVariableInfo() | Yes, even if previously busy | + * | SetVariable() | Yes, even if previously busy | + * | UpdateCapsule() | Yes, even if previously busy | + * | QueryCapsuleCapabilities() | Yes, even if previously busy | + * | ResetSystem() | Yes, even if previously busy | + * +----------------------------+------------------------------------------+ + * + * In order to prevent deadlocks under NMI, the wrappers for these functions + * may only grab the efi_runtime_lock or rtc_lock spinlocks if !efi_in_nmi(). + * However, not all of the services listed are reachable through NMI code paths, + * so the the special handling as suggested by the UEFI spec is only implemented + * for QueryVariableInfo() and SetVariable(), as these can be reached in NMI + * context through efi_pstore_write(). + */ +#ifndef efi_in_nmi +#define efi_in_nmi() (0) +#endif + /* * As per commit ef68c8f87ed1 ("x86: Serialize EFI time accesses on rtc_lock"), * the EFI specification requires that callers of the time related runtime @@ -32,7 +104,9 @@ static efi_status_t virt_efi_get_time(efi_time_t *tm, efi_time_cap_t *tc) efi_status_t status; spin_lock_irqsave(&rtc_lock, flags); + spin_lock(&efi_runtime_lock); status = efi_call_virt(get_time, tm, tc); + spin_unlock(&efi_runtime_lock); spin_unlock_irqrestore(&rtc_lock, flags); return status; } @@ -43,7 +117,9 @@ static efi_status_t virt_efi_set_time(efi_time_t *tm) efi_status_t status; spin_lock_irqsave(&rtc_lock, flags); + spin_lock(&efi_runtime_lock); status = efi_call_virt(set_time, tm); + spin_unlock(&efi_runtime_lock); spin_unlock_irqrestore(&rtc_lock, flags); return status; } @@ -56,7 +132,9 @@ static efi_status_t virt_efi_get_wakeup_time(efi_bool_t *enabled, efi_status_t status; spin_lock_irqsave(&rtc_lock, flags); + spin_lock(&efi_runtime_lock); status = efi_call_virt(get_wakeup_time, enabled, pending, tm); + spin_unlock(&efi_runtime_lock); spin_unlock_irqrestore(&rtc_lock, flags); return status; } @@ -67,7 +145,9 @@ static efi_status_t virt_efi_set_wakeup_time(efi_bool_t enabled, efi_time_t *tm) efi_status_t status; spin_lock_irqsave(&rtc_lock, flags); + spin_lock(&efi_runtime_lock); status = efi_call_virt(set_wakeup_time, enabled, tm); + spin_unlock(&efi_runtime_lock); spin_unlock_irqrestore(&rtc_lock, flags); return status; } @@ -78,14 +158,27 @@ static efi_status_t virt_efi_get_variable(efi_char16_t *name, unsigned long *data_size, void *data) { - return efi_call_virt(get_variable, name, vendor, attr, data_size, data); + unsigned long flags; + efi_status_t status; + + spin_lock_irqsave(&efi_runtime_lock, flags); + status = efi_call_virt(get_variable, name, vendor, attr, data_size, + data); + spin_unlock_irqrestore(&efi_runtime_lock, flags); + return status; } static efi_status_t virt_efi_get_next_variable(unsigned long *name_size, efi_char16_t *name, efi_guid_t *vendor) { - return efi_call_virt(get_next_variable, name_size, name, vendor); + unsigned long flags; + efi_status_t status; + + spin_lock_irqsave(&efi_runtime_lock, flags); + status = efi_call_virt(get_next_variable, name_size, name, vendor); + spin_unlock_irqrestore(&efi_runtime_lock, flags); + return status; } static efi_status_t virt_efi_set_variable(efi_char16_t *name, @@ -94,7 +187,17 @@ static efi_status_t virt_efi_set_variable(efi_char16_t *name, unsigned long data_size, void *data) { - return efi_call_virt(set_variable, name, vendor, attr, data_size, data); + unsigned long flags; + efi_status_t status; + bool __in_nmi = efi_in_nmi(); + + if (!__in_nmi) + spin_lock_irqsave(&efi_runtime_lock, flags); + status = efi_call_virt(set_variable, name, vendor, attr, data_size, + data); + if (!__in_nmi) + spin_unlock_irqrestore(&efi_runtime_lock, flags); + return status; } static efi_status_t virt_efi_query_variable_info(u32 attr, @@ -102,16 +205,31 @@ static efi_status_t virt_efi_query_variable_info(u32 attr, u64 *remaining_space, u64 *max_variable_size) { + unsigned long flags; + efi_status_t status; + bool __in_nmi = efi_in_nmi(); + if (efi.runtime_version < EFI_2_00_SYSTEM_TABLE_REVISION) return EFI_UNSUPPORTED; - return efi_call_virt(query_variable_info, attr, storage_space, - remaining_space, max_variable_size); + if (!__in_nmi) + spin_lock_irqsave(&efi_runtime_lock, flags); + status = efi_call_virt(query_variable_info, attr, storage_space, + remaining_space, max_variable_size); + if (!__in_nmi) + spin_unlock_irqrestore(&efi_runtime_lock, flags); + return status; } static efi_status_t virt_efi_get_next_high_mono_count(u32 *count) { - return efi_call_virt(get_next_high_mono_count, count); + unsigned long flags; + efi_status_t status; + + spin_lock_irqsave(&efi_runtime_lock, flags); + status = efi_call_virt(get_next_high_mono_count, count); + spin_unlock_irqrestore(&efi_runtime_lock, flags); + return status; } static void virt_efi_reset_system(int reset_type, @@ -119,17 +237,27 @@ static void virt_efi_reset_system(int reset_type, unsigned long data_size, efi_char16_t *data) { + unsigned long flags; + + spin_lock_irqsave(&efi_runtime_lock, flags); __efi_call_virt(reset_system, reset_type, status, data_size, data); + spin_unlock_irqrestore(&efi_runtime_lock, flags); } static efi_status_t virt_efi_update_capsule(efi_capsule_header_t **capsules, unsigned long count, unsigned long sg_list) { + unsigned long flags; + efi_status_t status; + if (efi.runtime_version < EFI_2_00_SYSTEM_TABLE_REVISION) return EFI_UNSUPPORTED; - return efi_call_virt(update_capsule, capsules, count, sg_list); + spin_lock_irqsave(&efi_runtime_lock, flags); + status = efi_call_virt(update_capsule, capsules, count, sg_list); + spin_unlock_irqrestore(&efi_runtime_lock, flags); + return status; } static efi_status_t virt_efi_query_capsule_caps(efi_capsule_header_t **capsules, @@ -137,11 +265,17 @@ static efi_status_t virt_efi_query_capsule_caps(efi_capsule_header_t **capsules, u64 *max_size, int *reset_type) { + unsigned long flags; + efi_status_t status; + if (efi.runtime_version < EFI_2_00_SYSTEM_TABLE_REVISION) return EFI_UNSUPPORTED; - return efi_call_virt(query_capsule_caps, capsules, count, max_size, - reset_type); + spin_lock_irqsave(&efi_runtime_lock, flags); + status = efi_call_virt(query_capsule_caps, capsules, count, max_size, + reset_type); + spin_unlock_irqrestore(&efi_runtime_lock, flags); + return status; } void efi_native_runtime_setup(void) -- cgit v1.2.3 From 6d80dba1c9fe4316ef626980102b92fa30c7845a Mon Sep 17 00:00:00 2001 From: Matt Fleming Date: Tue, 30 Sep 2014 21:58:52 +0100 Subject: efi: Provide a non-blocking SetVariable() operation There are some circumstances that call for trying to write an EFI variable in a non-blocking way. One such scenario is when writing pstore data in efi_pstore_write() via the pstore_dump() kdump callback. Now that we have an EFI runtime spinlock we need a way of aborting if there is contention instead of spinning, since when writing pstore data from the kdump callback, the runtime lock may already be held by the CPU that's running the callback if we crashed in the middle of an EFI variable operation. The situation is sufficiently special that a new EFI variable operation is warranted. Introduce ->set_variable_nonblocking() for this use case. It is an optional EFI backend operation, and need only be implemented by those backends that usually acquire locks to serialize access to EFI variables, as is the case for virt_efi_set_variable() where we now grab the EFI runtime spinlock. Cc: Peter Zijlstra Cc: Ingo Molnar Cc: Ard Biesheuvel Cc: Matthew Garrett Signed-off-by: Matt Fleming --- drivers/firmware/efi/runtime-wrappers.c | 19 +++++++++++++ drivers/firmware/efi/vars.c | 47 +++++++++++++++++++++++++++++++++ include/linux/efi.h | 6 +++++ 3 files changed, 72 insertions(+) (limited to 'drivers/firmware/efi/runtime-wrappers.c') diff --git a/drivers/firmware/efi/runtime-wrappers.c b/drivers/firmware/efi/runtime-wrappers.c index 9694cba665c4..4349206198b2 100644 --- a/drivers/firmware/efi/runtime-wrappers.c +++ b/drivers/firmware/efi/runtime-wrappers.c @@ -200,6 +200,24 @@ static efi_status_t virt_efi_set_variable(efi_char16_t *name, return status; } +static efi_status_t +virt_efi_set_variable_nonblocking(efi_char16_t *name, efi_guid_t *vendor, + u32 attr, unsigned long data_size, + void *data) +{ + unsigned long flags; + efi_status_t status; + + if (!spin_trylock_irqsave(&efi_runtime_lock, flags)) + return EFI_NOT_READY; + + status = efi_call_virt(set_variable, name, vendor, attr, data_size, + data); + spin_unlock_irqrestore(&efi_runtime_lock, flags); + return status; +} + + static efi_status_t virt_efi_query_variable_info(u32 attr, u64 *storage_space, u64 *remaining_space, @@ -287,6 +305,7 @@ void efi_native_runtime_setup(void) efi.get_variable = virt_efi_get_variable; efi.get_next_variable = virt_efi_get_next_variable; efi.set_variable = virt_efi_set_variable; + efi.set_variable_nonblocking = virt_efi_set_variable_nonblocking; efi.get_next_high_mono_count = virt_efi_get_next_high_mono_count; efi.reset_system = virt_efi_reset_system; efi.query_variable_info = virt_efi_query_variable_info; diff --git a/drivers/firmware/efi/vars.c b/drivers/firmware/efi/vars.c index 1fa724f31b0e..fa3c66bdc1e5 100644 --- a/drivers/firmware/efi/vars.c +++ b/drivers/firmware/efi/vars.c @@ -595,6 +595,39 @@ int efivar_entry_set(struct efivar_entry *entry, u32 attributes, } EXPORT_SYMBOL_GPL(efivar_entry_set); +/* + * efivar_entry_set_nonblocking - call set_variable_nonblocking() + * + * This function is guaranteed to not block and is suitable for calling + * from crash/panic handlers. + * + * Crucially, this function will not block if it cannot acquire + * __efivars->lock. Instead, it returns -EBUSY. + */ +static int +efivar_entry_set_nonblocking(efi_char16_t *name, efi_guid_t vendor, + u32 attributes, unsigned long size, void *data) +{ + const struct efivar_operations *ops = __efivars->ops; + unsigned long flags; + efi_status_t status; + + if (!spin_trylock_irqsave(&__efivars->lock, flags)) + return -EBUSY; + + status = check_var_size(attributes, size + ucs2_strsize(name, 1024)); + if (status != EFI_SUCCESS) { + spin_unlock_irqrestore(&__efivars->lock, flags); + return -ENOSPC; + } + + status = ops->set_variable_nonblocking(name, &vendor, attributes, + size, data); + + spin_unlock_irqrestore(&__efivars->lock, flags); + return efi_status_to_err(status); +} + /** * efivar_entry_set_safe - call set_variable() if enough space in firmware * @name: buffer containing the variable name @@ -622,6 +655,20 @@ int efivar_entry_set_safe(efi_char16_t *name, efi_guid_t vendor, u32 attributes, if (!ops->query_variable_store) return -ENOSYS; + /* + * If the EFI variable backend provides a non-blocking + * ->set_variable() operation and we're in a context where we + * cannot block, then we need to use it to avoid live-locks, + * since the implication is that the regular ->set_variable() + * will block. + * + * If no ->set_variable_nonblocking() is provided then + * ->set_variable() is assumed to be non-blocking. + */ + if (!block && ops->set_variable_nonblocking) + return efivar_entry_set_nonblocking(name, vendor, attributes, + size, data); + if (!block) { if (!spin_trylock_irqsave(&__efivars->lock, flags)) return -EBUSY; diff --git a/include/linux/efi.h b/include/linux/efi.h index 78b29b133e14..0949f9c7e872 100644 --- a/include/linux/efi.h +++ b/include/linux/efi.h @@ -503,6 +503,10 @@ typedef efi_status_t efi_get_next_variable_t (unsigned long *name_size, efi_char typedef efi_status_t efi_set_variable_t (efi_char16_t *name, efi_guid_t *vendor, u32 attr, unsigned long data_size, void *data); +typedef efi_status_t +efi_set_variable_nonblocking_t(efi_char16_t *name, efi_guid_t *vendor, + u32 attr, unsigned long data_size, void *data); + typedef efi_status_t efi_get_next_high_mono_count_t (u32 *count); typedef void efi_reset_system_t (int reset_type, efi_status_t status, unsigned long data_size, efi_char16_t *data); @@ -822,6 +826,7 @@ extern struct efi { efi_get_variable_t *get_variable; efi_get_next_variable_t *get_next_variable; efi_set_variable_t *set_variable; + efi_set_variable_nonblocking_t *set_variable_nonblocking; efi_query_variable_info_t *query_variable_info; efi_update_capsule_t *update_capsule; efi_query_capsule_caps_t *query_capsule_caps; @@ -1042,6 +1047,7 @@ struct efivar_operations { efi_get_variable_t *get_variable; efi_get_next_variable_t *get_next_variable; efi_set_variable_t *set_variable; + efi_set_variable_nonblocking_t *set_variable_nonblocking; efi_query_variable_store_t *query_variable_store; }; -- cgit v1.2.3 From 60b4dc7720a5251f5dbda69ad404e0bcb3cb6bfb Mon Sep 17 00:00:00 2001 From: Matt Fleming Date: Tue, 30 Sep 2014 15:03:38 +0100 Subject: efi: Delete the in_nmi() conditional runtime locking commit 5dc3826d9f08 ("efi: Implement mandatory locking for UEFI Runtime Services") implemented some conditional locking when accessing variable runtime services that Ingo described as "pretty disgusting". The intention with the !efi_in_nmi() checks was to avoid live-locks when trying to write pstore crash data into an EFI variable. Such lockless accesses are allowed according to the UEFI specification when we're in a "non-recoverable" state, but whether or not things are implemented correctly in actual firmware implementations remains an unanswered question, and so it would seem sensible to avoid doing any kind of unsynchronized variable accesses. Furthermore, the efi_in_nmi() tests are inadequate because they don't account for the case where we call EFI variable services from panic or oops callbacks and aren't executing in NMI context. In other words, live-locking is still possible. Let's just remove the conditional locking altogether. Now we've got the ->set_variable_nonblocking() EFI variable operation we can abort if the runtime lock is already held. Aborting is by far the safest option. Cc: Peter Zijlstra Cc: Ingo Molnar Cc: Ard Biesheuvel Cc: Matthew Garrett Signed-off-by: Matt Fleming --- arch/x86/include/asm/efi.h | 2 -- drivers/firmware/efi/runtime-wrappers.c | 17 ++++------------- 2 files changed, 4 insertions(+), 15 deletions(-) (limited to 'drivers/firmware/efi/runtime-wrappers.c') diff --git a/arch/x86/include/asm/efi.h b/arch/x86/include/asm/efi.h index abba1fe1db5e..408ad6d3222e 100644 --- a/arch/x86/include/asm/efi.h +++ b/arch/x86/include/asm/efi.h @@ -86,8 +86,6 @@ extern void __iomem *__init efi_ioremap(unsigned long addr, unsigned long size, #endif /* CONFIG_X86_32 */ -#define efi_in_nmi() in_nmi() - extern struct efi_scratch efi_scratch; extern void __init efi_set_executable(efi_memory_desc_t *md, bool executable); extern int __init efi_memblock_x86_reserve_range(void); diff --git a/drivers/firmware/efi/runtime-wrappers.c b/drivers/firmware/efi/runtime-wrappers.c index 4349206198b2..228bbf910461 100644 --- a/drivers/firmware/efi/runtime-wrappers.c +++ b/drivers/firmware/efi/runtime-wrappers.c @@ -86,9 +86,6 @@ static DEFINE_SPINLOCK(efi_runtime_lock); * for QueryVariableInfo() and SetVariable(), as these can be reached in NMI * context through efi_pstore_write(). */ -#ifndef efi_in_nmi -#define efi_in_nmi() (0) -#endif /* * As per commit ef68c8f87ed1 ("x86: Serialize EFI time accesses on rtc_lock"), @@ -189,14 +186,11 @@ static efi_status_t virt_efi_set_variable(efi_char16_t *name, { unsigned long flags; efi_status_t status; - bool __in_nmi = efi_in_nmi(); - if (!__in_nmi) - spin_lock_irqsave(&efi_runtime_lock, flags); + spin_lock_irqsave(&efi_runtime_lock, flags); status = efi_call_virt(set_variable, name, vendor, attr, data_size, data); - if (!__in_nmi) - spin_unlock_irqrestore(&efi_runtime_lock, flags); + spin_unlock_irqrestore(&efi_runtime_lock, flags); return status; } @@ -225,17 +219,14 @@ static efi_status_t virt_efi_query_variable_info(u32 attr, { unsigned long flags; efi_status_t status; - bool __in_nmi = efi_in_nmi(); if (efi.runtime_version < EFI_2_00_SYSTEM_TABLE_REVISION) return EFI_UNSUPPORTED; - if (!__in_nmi) - spin_lock_irqsave(&efi_runtime_lock, flags); + spin_lock_irqsave(&efi_runtime_lock, flags); status = efi_call_virt(query_variable_info, attr, storage_space, remaining_space, max_variable_size); - if (!__in_nmi) - spin_unlock_irqrestore(&efi_runtime_lock, flags); + spin_unlock_irqrestore(&efi_runtime_lock, flags); return status; } -- cgit v1.2.3