| Message ID | 20180711170426.11133-1-ard.biesheuvel@linaro.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Wed, Jul 11, 2018 at 10:04 AM, Ard Biesheuvel <ard.biesheuvel@linaro.org> wrote: > When CONFIG_STRICT_KERNEL_RWX=y [which is the default on arm64], we > take great care to ensure that the mappings of modules in the vmalloc > space are locked down as much as possible, i.e., executable code is > mapped read-only, read-only data is mapped read-only and non-executable, > and read-write data is mapped non-executable as well. > > However, due to the way we map the linear region [aka the kernel direct > mapping], those module regions are aliased by read-write mappings, and > it is possible for an attacker to modify code or data that was assumed > to be immutable in this configuration. > > So let's ensure that the linear alias of module memory is remapped > read-only upon allocation and remapped read-write when it is freed. The > latter requires some special handling involving a workqueue due to the > fact that it may be called in softirq context at which time calling > find_vm_area() is unsafe. > > Note that this requires the entire linear region to be mapped down to > pages, which may result in a performance regression in some hardware > configurations. So this is only module allocations, not all vmap allocations? This should cover modules and eBPF, though any future stuff that uses vmalloc and wants to go read-only made need tweaking. > > Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> > --- > Changes since RFC/v1: > - remap linear alias read-only rather than invalid > - honour rodata_enabled, i.e., 'rodata=off' will disable this functionality > - use VMA nr_pages field rather than size to obtain the number of pages to > remap > > arch/arm64/include/asm/module.h | 6 +++ > arch/arm64/kernel/module.c | 42 ++++++++++++++++++++ > arch/arm64/mm/mmu.c | 2 +- > arch/arm64/mm/pageattr.c | 19 +++++++++ > 4 files changed, 68 insertions(+), 1 deletion(-) > > diff --git a/arch/arm64/include/asm/module.h b/arch/arm64/include/asm/module.h > index 97d0ef12e2ff..f16f0073642b 100644 > --- a/arch/arm64/include/asm/module.h > +++ b/arch/arm64/include/asm/module.h > @@ -91,4 +91,10 @@ static inline bool plt_entries_equal(const struct plt_entry *a, > a->mov2 == b->mov2; > } > > +#ifdef CONFIG_STRICT_KERNEL_RWX > +void remap_linear_module_alias(void *module_region, bool ro); > +#else > +static inline void remap_linear_module_alias(void *module_region, bool ro) {} > +#endif > + > #endif /* __ASM_MODULE_H */ > diff --git a/arch/arm64/kernel/module.c b/arch/arm64/kernel/module.c > index f0f27aeefb73..2b61ca0285eb 100644 > --- a/arch/arm64/kernel/module.c > +++ b/arch/arm64/kernel/module.c > @@ -26,10 +26,51 @@ > #include <linux/mm.h> > #include <linux/moduleloader.h> > #include <linux/vmalloc.h> > +#include <linux/workqueue.h> > #include <asm/alternative.h> > #include <asm/insn.h> > #include <asm/sections.h> > > +#ifdef CONFIG_STRICT_KERNEL_RWX > + > +static struct workqueue_struct *module_free_wq; > + > +static int init_workqueue(void) > +{ > + module_free_wq = alloc_ordered_workqueue("module_free_wq", 0); > + WARN_ON(!module_free_wq); > + > + return 0; > +} > +pure_initcall(init_workqueue); > + > +static void module_free_wq_worker(struct work_struct *work) > +{ > + remap_linear_module_alias(work, false); > + vfree(work); > +} > + > +void module_memfree(void *module_region) > +{ > + struct work_struct *work; > + > + if (!module_region) > + return; > + > + /* > + * At this point, module_region is a pointer to an allocation of at > + * least PAGE_SIZE bytes that is mapped read-write. So instead of > + * allocating memory for a data structure containing a work_struct > + * instance and a copy of the value of module_region, just reuse the > + * allocation directly. > + */ This took me a few reads to realize this is the virtual memory address, not the linear address -- duh. I was thinking "but it's not read-write yet!" :) > + work = module_region; > + INIT_WORK(work, module_free_wq_worker); > + queue_work(module_free_wq, work); > +} > + > +#endif > + > void *module_alloc(unsigned long size) > { > gfp_t gfp_mask = GFP_KERNEL; > @@ -65,6 +106,7 @@ void *module_alloc(unsigned long size) > return NULL; > } > > + remap_linear_module_alias(p, true); > return p; > } > > diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c > index 493ff75670ff..5492b691aafd 100644 > --- a/arch/arm64/mm/mmu.c > +++ b/arch/arm64/mm/mmu.c > @@ -432,7 +432,7 @@ static void __init map_mem(pgd_t *pgdp) > struct memblock_region *reg; > int flags = 0; > > - if (debug_pagealloc_enabled()) > + if (rodata_enabled || debug_pagealloc_enabled()) > flags = NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS; Is it worth adding a comment above this to describe what the conditions are that are triggering the page-level granularity here? > > /* > diff --git a/arch/arm64/mm/pageattr.c b/arch/arm64/mm/pageattr.c > index a56359373d8b..cc04be572660 100644 > --- a/arch/arm64/mm/pageattr.c > +++ b/arch/arm64/mm/pageattr.c > @@ -138,6 +138,25 @@ int set_memory_valid(unsigned long addr, int numpages, int enable) > __pgprot(PTE_VALID)); > } > > +#ifdef CONFIG_STRICT_KERNEL_RWX > +void remap_linear_module_alias(void *module_region, bool ro) > +{ > + struct vm_struct *vm; > + int i; > + > + if (!rodata_enabled) > + return; > + > + vm = find_vm_area(module_region); > + WARN_ON(!vm || !vm->pages); > + > + for (i = 0; i < vm->nr_pages; i++) > + __change_memory_common((u64)page_address(vm->pages[i]), PAGE_SIZE, > + ro ? __pgprot(PTE_RDONLY) : __pgprot(PTE_WRITE), > + ro ? __pgprot(PTE_WRITE) : __pgprot(PTE_RDONLY)); > +} > +#endif > + > #ifdef CONFIG_DEBUG_PAGEALLOC > void __kernel_map_pages(struct page *page, int numpages, int enable) > { > -- > 2.17.1 > If this doesn't get another version, probably just skip my comment suggestion above. :) Regardless: Reviewed-by: Kees Cook <keescook@chromium.org> So glad to have the linear mapping handled now. :) I should probably add tests for this to lkdtm, since it's a per-architecture thing... -Kees
On 07/11/2018 10:04 AM, Ard Biesheuvel wrote: > When CONFIG_STRICT_KERNEL_RWX=y [which is the default on arm64], we > take great care to ensure that the mappings of modules in the vmalloc > space are locked down as much as possible, i.e., executable code is > mapped read-only, read-only data is mapped read-only and non-executable, > and read-write data is mapped non-executable as well. > > However, due to the way we map the linear region [aka the kernel direct > mapping], those module regions are aliased by read-write mappings, and > it is possible for an attacker to modify code or data that was assumed > to be immutable in this configuration. > > So let's ensure that the linear alias of module memory is remapped > read-only upon allocation and remapped read-write when it is freed. The > latter requires some special handling involving a workqueue due to the > fact that it may be called in softirq context at which time calling > find_vm_area() is unsafe. > > Note that this requires the entire linear region to be mapped down to > pages, which may result in a performance regression in some hardware > configurations. > I'm concerned any performance regression might just make people turn off rodata all together. It's not overly bad for the kernel mappings but modules are fully RWX so we might be trading a subtle threat for a more obvious one. Would putting this behind a Kconfig make sense for those who want the status quo (with a big caveat that this should have been present from the start)? Thanks, Laura > Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> > --- > Changes since RFC/v1: > - remap linear alias read-only rather than invalid > - honour rodata_enabled, i.e., 'rodata=off' will disable this functionality > - use VMA nr_pages field rather than size to obtain the number of pages to > remap > > arch/arm64/include/asm/module.h | 6 +++ > arch/arm64/kernel/module.c | 42 ++++++++++++++++++++ > arch/arm64/mm/mmu.c | 2 +- > arch/arm64/mm/pageattr.c | 19 +++++++++ > 4 files changed, 68 insertions(+), 1 deletion(-) > > diff --git a/arch/arm64/include/asm/module.h b/arch/arm64/include/asm/module.h > index 97d0ef12e2ff..f16f0073642b 100644 > --- a/arch/arm64/include/asm/module.h > +++ b/arch/arm64/include/asm/module.h > @@ -91,4 +91,10 @@ static inline bool plt_entries_equal(const struct plt_entry *a, > a->mov2 == b->mov2; > } > > +#ifdef CONFIG_STRICT_KERNEL_RWX > +void remap_linear_module_alias(void *module_region, bool ro); > +#else > +static inline void remap_linear_module_alias(void *module_region, bool ro) {} > +#endif > + > #endif /* __ASM_MODULE_H */ > diff --git a/arch/arm64/kernel/module.c b/arch/arm64/kernel/module.c > index f0f27aeefb73..2b61ca0285eb 100644 > --- a/arch/arm64/kernel/module.c > +++ b/arch/arm64/kernel/module.c > @@ -26,10 +26,51 @@ > #include <linux/mm.h> > #include <linux/moduleloader.h> > #include <linux/vmalloc.h> > +#include <linux/workqueue.h> > #include <asm/alternative.h> > #include <asm/insn.h> > #include <asm/sections.h> > > +#ifdef CONFIG_STRICT_KERNEL_RWX > + > +static struct workqueue_struct *module_free_wq; > + > +static int init_workqueue(void) > +{ > + module_free_wq = alloc_ordered_workqueue("module_free_wq", 0); > + WARN_ON(!module_free_wq); > + > + return 0; > +} > +pure_initcall(init_workqueue); > + > +static void module_free_wq_worker(struct work_struct *work) > +{ > + remap_linear_module_alias(work, false); > + vfree(work); > +} > + > +void module_memfree(void *module_region) > +{ > + struct work_struct *work; > + > + if (!module_region) > + return; > + > + /* > + * At this point, module_region is a pointer to an allocation of at > + * least PAGE_SIZE bytes that is mapped read-write. So instead of > + * allocating memory for a data structure containing a work_struct > + * instance and a copy of the value of module_region, just reuse the > + * allocation directly. > + */ > + work = module_region; > + INIT_WORK(work, module_free_wq_worker); > + queue_work(module_free_wq, work); > +} > + > +#endif > + > void *module_alloc(unsigned long size) > { > gfp_t gfp_mask = GFP_KERNEL; > @@ -65,6 +106,7 @@ void *module_alloc(unsigned long size) > return NULL; > } > > + remap_linear_module_alias(p, true); > return p; > } > > diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c > index 493ff75670ff..5492b691aafd 100644 > --- a/arch/arm64/mm/mmu.c > +++ b/arch/arm64/mm/mmu.c > @@ -432,7 +432,7 @@ static void __init map_mem(pgd_t *pgdp) > struct memblock_region *reg; > int flags = 0; > > - if (debug_pagealloc_enabled()) > + if (rodata_enabled || debug_pagealloc_enabled()) > flags = NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS; > > /* > diff --git a/arch/arm64/mm/pageattr.c b/arch/arm64/mm/pageattr.c > index a56359373d8b..cc04be572660 100644 > --- a/arch/arm64/mm/pageattr.c > +++ b/arch/arm64/mm/pageattr.c > @@ -138,6 +138,25 @@ int set_memory_valid(unsigned long addr, int numpages, int enable) > __pgprot(PTE_VALID)); > } > > +#ifdef CONFIG_STRICT_KERNEL_RWX > +void remap_linear_module_alias(void *module_region, bool ro) > +{ > + struct vm_struct *vm; > + int i; > + > + if (!rodata_enabled) > + return; > + > + vm = find_vm_area(module_region); > + WARN_ON(!vm || !vm->pages); > + > + for (i = 0; i < vm->nr_pages; i++) > + __change_memory_common((u64)page_address(vm->pages[i]), PAGE_SIZE, > + ro ? __pgprot(PTE_RDONLY) : __pgprot(PTE_WRITE), > + ro ? __pgprot(PTE_WRITE) : __pgprot(PTE_RDONLY)); > +} > +#endif > + > #ifdef CONFIG_DEBUG_PAGEALLOC > void __kernel_map_pages(struct page *page, int numpages, int enable) > { >
On 17 July 2018 at 07:56, Laura Abbott <labbott@redhat.com> wrote: > On 07/11/2018 10:04 AM, Ard Biesheuvel wrote: >> >> When CONFIG_STRICT_KERNEL_RWX=y [which is the default on arm64], we >> take great care to ensure that the mappings of modules in the vmalloc >> space are locked down as much as possible, i.e., executable code is >> mapped read-only, read-only data is mapped read-only and non-executable, >> and read-write data is mapped non-executable as well. >> >> However, due to the way we map the linear region [aka the kernel direct >> mapping], those module regions are aliased by read-write mappings, and >> it is possible for an attacker to modify code or data that was assumed >> to be immutable in this configuration. >> >> So let's ensure that the linear alias of module memory is remapped >> read-only upon allocation and remapped read-write when it is freed. The >> latter requires some special handling involving a workqueue due to the >> fact that it may be called in softirq context at which time calling >> find_vm_area() is unsafe. >> >> Note that this requires the entire linear region to be mapped down to >> pages, which may result in a performance regression in some hardware >> configurations. >> > > I'm concerned any performance regression might just make people turn > off rodata all together. It's not overly bad for the kernel mappings > but modules are fully RWX so we might be trading a subtle threat for > a more obvious one. Would putting this behind a Kconfig make sense > for those who want the status quo (with a big caveat that this should > have been present from the start)? > It would be good if we could quantify the performance hit before deciding that it is a problem. I was actually expecting more opposition and/or debate anyway in response to this patch, along the lines of 'use case X is Y% slower now so your patch sucks!' but sadly, none of that yet.
Hi Ard, Remember writing this patch? ;) On Wed, Jul 11, 2018 at 07:04:26PM +0200, Ard Biesheuvel wrote: > When CONFIG_STRICT_KERNEL_RWX=y [which is the default on arm64], we > take great care to ensure that the mappings of modules in the vmalloc > space are locked down as much as possible, i.e., executable code is > mapped read-only, read-only data is mapped read-only and non-executable, > and read-write data is mapped non-executable as well. > > However, due to the way we map the linear region [aka the kernel direct > mapping], those module regions are aliased by read-write mappings, and > it is possible for an attacker to modify code or data that was assumed > to be immutable in this configuration. > > So let's ensure that the linear alias of module memory is remapped > read-only upon allocation and remapped read-write when it is freed. The > latter requires some special handling involving a workqueue due to the > fact that it may be called in softirq context at which time calling > find_vm_area() is unsafe. > > Note that this requires the entire linear region to be mapped down to > pages, which may result in a performance regression in some hardware > configurations. > > Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> > --- > Changes since RFC/v1: > - remap linear alias read-only rather than invalid > - honour rodata_enabled, i.e., 'rodata=off' will disable this functionality > - use VMA nr_pages field rather than size to obtain the number of pages to > remap > > arch/arm64/include/asm/module.h | 6 +++ > arch/arm64/kernel/module.c | 42 ++++++++++++++++++++ > arch/arm64/mm/mmu.c | 2 +- > arch/arm64/mm/pageattr.c | 19 +++++++++ > 4 files changed, 68 insertions(+), 1 deletion(-) > > diff --git a/arch/arm64/include/asm/module.h b/arch/arm64/include/asm/module.h > index 97d0ef12e2ff..f16f0073642b 100644 > --- a/arch/arm64/include/asm/module.h > +++ b/arch/arm64/include/asm/module.h > @@ -91,4 +91,10 @@ static inline bool plt_entries_equal(const struct plt_entry *a, > a->mov2 == b->mov2; > } > > +#ifdef CONFIG_STRICT_KERNEL_RWX > +void remap_linear_module_alias(void *module_region, bool ro); > +#else > +static inline void remap_linear_module_alias(void *module_region, bool ro) {} > +#endif > + > #endif /* __ASM_MODULE_H */ > diff --git a/arch/arm64/kernel/module.c b/arch/arm64/kernel/module.c > index f0f27aeefb73..2b61ca0285eb 100644 > --- a/arch/arm64/kernel/module.c > +++ b/arch/arm64/kernel/module.c > @@ -26,10 +26,51 @@ > #include <linux/mm.h> > #include <linux/moduleloader.h> > #include <linux/vmalloc.h> > +#include <linux/workqueue.h> > #include <asm/alternative.h> > #include <asm/insn.h> > #include <asm/sections.h> > > +#ifdef CONFIG_STRICT_KERNEL_RWX > + > +static struct workqueue_struct *module_free_wq; > + > +static int init_workqueue(void) > +{ > + module_free_wq = alloc_ordered_workqueue("module_free_wq", 0); > + WARN_ON(!module_free_wq); > + > + return 0; > +} > +pure_initcall(init_workqueue); > + > +static void module_free_wq_worker(struct work_struct *work) > +{ > + remap_linear_module_alias(work, false); > + vfree(work); > +} > + > +void module_memfree(void *module_region) > +{ > + struct work_struct *work; > + > + if (!module_region) > + return; > + > + /* > + * At this point, module_region is a pointer to an allocation of at > + * least PAGE_SIZE bytes that is mapped read-write. So instead of > + * allocating memory for a data structure containing a work_struct > + * instance and a copy of the value of module_region, just reuse the > + * allocation directly. > + */ > + work = module_region; > + INIT_WORK(work, module_free_wq_worker); > + queue_work(module_free_wq, work); > +} Whilst I can see that this works, it's a bit grotty because we're duplicating much of the hoop-jumping that vfree() also has to go through. Given that we allocate the module memory in the arch code, could we grab the pages at alloc time and stash them in the mod_arch_specific field? > +#endif > + > void *module_alloc(unsigned long size) > { > gfp_t gfp_mask = GFP_KERNEL; > @@ -65,6 +106,7 @@ void *module_alloc(unsigned long size) > return NULL; > } > > + remap_linear_module_alias(p, true); > return p; > } > > diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c > index 493ff75670ff..5492b691aafd 100644 > --- a/arch/arm64/mm/mmu.c > +++ b/arch/arm64/mm/mmu.c > @@ -432,7 +432,7 @@ static void __init map_mem(pgd_t *pgdp) > struct memblock_region *reg; > int flags = 0; > > - if (debug_pagealloc_enabled()) > + if (rodata_enabled || debug_pagealloc_enabled()) > flags = NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS; I suppose this is the most contentious part of the patch. Perhaps a better solution would be to tweak CONFIG_STRICT_MODULE_RWX so that it's structured more like CONFIG_STACKPROTECTOR. That way, your patch here could be the "strong" version, which comes with more of a performance cost. > diff --git a/arch/arm64/mm/pageattr.c b/arch/arm64/mm/pageattr.c > index a56359373d8b..cc04be572660 100644 > --- a/arch/arm64/mm/pageattr.c > +++ b/arch/arm64/mm/pageattr.c > @@ -138,6 +138,25 @@ int set_memory_valid(unsigned long addr, int numpages, int enable) > __pgprot(PTE_VALID)); > } > > +#ifdef CONFIG_STRICT_KERNEL_RWX CONFIG_STRICT_MODULE_RWX instead? > +void remap_linear_module_alias(void *module_region, bool ro) > +{ > + struct vm_struct *vm; > + int i; > + > + if (!rodata_enabled) > + return; > + > + vm = find_vm_area(module_region); > + WARN_ON(!vm || !vm->pages); > + > + for (i = 0; i < vm->nr_pages; i++) > + __change_memory_common((u64)page_address(vm->pages[i]), PAGE_SIZE, > + ro ? __pgprot(PTE_RDONLY) : __pgprot(PTE_WRITE), > + ro ? __pgprot(PTE_WRITE) : __pgprot(PTE_RDONLY)); > +} > +#endif I think it would be cleaner just to have two wrapper functions for this: remap_linear_module_alias_ro remap_linear_module_alias_rw so we can drop the bool parameter from the API and just pass the set/clear attributes directly to something like __remap_linear_module_alias. Will
On 1 November 2018 at 16:17, Will Deacon <will.deacon@arm.com> wrote: > Hi Ard, > > Remember writing this patch? ;) > > On Wed, Jul 11, 2018 at 07:04:26PM +0200, Ard Biesheuvel wrote: >> When CONFIG_STRICT_KERNEL_RWX=y [which is the default on arm64], we >> take great care to ensure that the mappings of modules in the vmalloc >> space are locked down as much as possible, i.e., executable code is >> mapped read-only, read-only data is mapped read-only and non-executable, >> and read-write data is mapped non-executable as well. >> >> However, due to the way we map the linear region [aka the kernel direct >> mapping], those module regions are aliased by read-write mappings, and >> it is possible for an attacker to modify code or data that was assumed >> to be immutable in this configuration. >> >> So let's ensure that the linear alias of module memory is remapped >> read-only upon allocation and remapped read-write when it is freed. The >> latter requires some special handling involving a workqueue due to the >> fact that it may be called in softirq context at which time calling >> find_vm_area() is unsafe. >> >> Note that this requires the entire linear region to be mapped down to >> pages, which may result in a performance regression in some hardware >> configurations. >> >> Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> >> --- >> Changes since RFC/v1: >> - remap linear alias read-only rather than invalid >> - honour rodata_enabled, i.e., 'rodata=off' will disable this functionality >> - use VMA nr_pages field rather than size to obtain the number of pages to >> remap >> >> arch/arm64/include/asm/module.h | 6 +++ >> arch/arm64/kernel/module.c | 42 ++++++++++++++++++++ >> arch/arm64/mm/mmu.c | 2 +- >> arch/arm64/mm/pageattr.c | 19 +++++++++ >> 4 files changed, 68 insertions(+), 1 deletion(-) >> >> diff --git a/arch/arm64/include/asm/module.h b/arch/arm64/include/asm/module.h >> index 97d0ef12e2ff..f16f0073642b 100644 >> --- a/arch/arm64/include/asm/module.h >> +++ b/arch/arm64/include/asm/module.h >> @@ -91,4 +91,10 @@ static inline bool plt_entries_equal(const struct plt_entry *a, >> a->mov2 == b->mov2; >> } >> >> +#ifdef CONFIG_STRICT_KERNEL_RWX >> +void remap_linear_module_alias(void *module_region, bool ro); >> +#else >> +static inline void remap_linear_module_alias(void *module_region, bool ro) {} >> +#endif >> + >> #endif /* __ASM_MODULE_H */ >> diff --git a/arch/arm64/kernel/module.c b/arch/arm64/kernel/module.c >> index f0f27aeefb73..2b61ca0285eb 100644 >> --- a/arch/arm64/kernel/module.c >> +++ b/arch/arm64/kernel/module.c >> @@ -26,10 +26,51 @@ >> #include <linux/mm.h> >> #include <linux/moduleloader.h> >> #include <linux/vmalloc.h> >> +#include <linux/workqueue.h> >> #include <asm/alternative.h> >> #include <asm/insn.h> >> #include <asm/sections.h> >> >> +#ifdef CONFIG_STRICT_KERNEL_RWX >> + >> +static struct workqueue_struct *module_free_wq; >> + >> +static int init_workqueue(void) >> +{ >> + module_free_wq = alloc_ordered_workqueue("module_free_wq", 0); >> + WARN_ON(!module_free_wq); >> + >> + return 0; >> +} >> +pure_initcall(init_workqueue); >> + >> +static void module_free_wq_worker(struct work_struct *work) >> +{ >> + remap_linear_module_alias(work, false); >> + vfree(work); >> +} >> + >> +void module_memfree(void *module_region) >> +{ >> + struct work_struct *work; >> + >> + if (!module_region) >> + return; >> + >> + /* >> + * At this point, module_region is a pointer to an allocation of at >> + * least PAGE_SIZE bytes that is mapped read-write. So instead of >> + * allocating memory for a data structure containing a work_struct >> + * instance and a copy of the value of module_region, just reuse the >> + * allocation directly. >> + */ >> + work = module_region; >> + INIT_WORK(work, module_free_wq_worker); >> + queue_work(module_free_wq, work); >> +} > > Whilst I can see that this works, it's a bit grotty because we're > duplicating much of the hoop-jumping that vfree() also has to go through. > Given that we allocate the module memory in the arch code, could we grab > the pages at alloc time and stash them in the mod_arch_specific field? > Yeah that should work >> +#endif >> + >> void *module_alloc(unsigned long size) >> { >> gfp_t gfp_mask = GFP_KERNEL; >> @@ -65,6 +106,7 @@ void *module_alloc(unsigned long size) >> return NULL; >> } >> >> + remap_linear_module_alias(p, true); >> return p; >> } >> >> diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c >> index 493ff75670ff..5492b691aafd 100644 >> --- a/arch/arm64/mm/mmu.c >> +++ b/arch/arm64/mm/mmu.c >> @@ -432,7 +432,7 @@ static void __init map_mem(pgd_t *pgdp) >> struct memblock_region *reg; >> int flags = 0; >> >> - if (debug_pagealloc_enabled()) >> + if (rodata_enabled || debug_pagealloc_enabled()) >> flags = NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS; > > I suppose this is the most contentious part of the patch. Perhaps a > better solution would be to tweak CONFIG_STRICT_MODULE_RWX so that it's > structured more like CONFIG_STACKPROTECTOR. That way, your patch here > could be the "strong" version, which comes with more of a performance > cost. > I'd like to understand whether this performance hit is theoretical or not. It is obviously less efficient to map the linear region at page granularity, but which workloads will actually notice the difference? Also, I think this should be a runtime setting, perhaps rodata=full? With a CONFIG_ option to set the default. >> diff --git a/arch/arm64/mm/pageattr.c b/arch/arm64/mm/pageattr.c >> index a56359373d8b..cc04be572660 100644 >> --- a/arch/arm64/mm/pageattr.c >> +++ b/arch/arm64/mm/pageattr.c >> @@ -138,6 +138,25 @@ int set_memory_valid(unsigned long addr, int numpages, int enable) >> __pgprot(PTE_VALID)); >> } >> >> +#ifdef CONFIG_STRICT_KERNEL_RWX > > CONFIG_STRICT_MODULE_RWX instead? > Yes. >> +void remap_linear_module_alias(void *module_region, bool ro) >> +{ >> + struct vm_struct *vm; >> + int i; >> + >> + if (!rodata_enabled) >> + return; >> + >> + vm = find_vm_area(module_region); >> + WARN_ON(!vm || !vm->pages); >> + >> + for (i = 0; i < vm->nr_pages; i++) >> + __change_memory_common((u64)page_address(vm->pages[i]), PAGE_SIZE, >> + ro ? __pgprot(PTE_RDONLY) : __pgprot(PTE_WRITE), >> + ro ? __pgprot(PTE_WRITE) : __pgprot(PTE_RDONLY)); >> +} >> +#endif > > I think it would be cleaner just to have two wrapper functions for this: > > remap_linear_module_alias_ro > remap_linear_module_alias_rw > > so we can drop the bool parameter from the API and just pass the set/clear > attributes directly to something like __remap_linear_module_alias. > Fair enough.
On Mon, Nov 05, 2018 at 01:29:34PM +0100, Ard Biesheuvel wrote: > On 1 November 2018 at 16:17, Will Deacon <will.deacon@arm.com> wrote: > > On Wed, Jul 11, 2018 at 07:04:26PM +0200, Ard Biesheuvel wrote: > >> diff --git a/arch/arm64/include/asm/module.h b/arch/arm64/include/asm/module.h > >> index 97d0ef12e2ff..f16f0073642b 100644 > >> --- a/arch/arm64/include/asm/module.h > >> +++ b/arch/arm64/include/asm/module.h > >> @@ -91,4 +91,10 @@ static inline bool plt_entries_equal(const struct plt_entry *a, > >> a->mov2 == b->mov2; > >> } > >> > >> +#ifdef CONFIG_STRICT_KERNEL_RWX > >> +void remap_linear_module_alias(void *module_region, bool ro); > >> +#else > >> +static inline void remap_linear_module_alias(void *module_region, bool ro) {} > >> +#endif > >> + > >> #endif /* __ASM_MODULE_H */ > >> diff --git a/arch/arm64/kernel/module.c b/arch/arm64/kernel/module.c > >> index f0f27aeefb73..2b61ca0285eb 100644 > >> --- a/arch/arm64/kernel/module.c > >> +++ b/arch/arm64/kernel/module.c > >> @@ -26,10 +26,51 @@ > >> #include <linux/mm.h> > >> #include <linux/moduleloader.h> > >> #include <linux/vmalloc.h> > >> +#include <linux/workqueue.h> > >> #include <asm/alternative.h> > >> #include <asm/insn.h> > >> #include <asm/sections.h> > >> > >> +#ifdef CONFIG_STRICT_KERNEL_RWX > >> + > >> +static struct workqueue_struct *module_free_wq; > >> + > >> +static int init_workqueue(void) > >> +{ > >> + module_free_wq = alloc_ordered_workqueue("module_free_wq", 0); > >> + WARN_ON(!module_free_wq); > >> + > >> + return 0; > >> +} > >> +pure_initcall(init_workqueue); > >> + > >> +static void module_free_wq_worker(struct work_struct *work) > >> +{ > >> + remap_linear_module_alias(work, false); > >> + vfree(work); > >> +} > >> + > >> +void module_memfree(void *module_region) > >> +{ > >> + struct work_struct *work; > >> + > >> + if (!module_region) > >> + return; > >> + > >> + /* > >> + * At this point, module_region is a pointer to an allocation of at > >> + * least PAGE_SIZE bytes that is mapped read-write. So instead of > >> + * allocating memory for a data structure containing a work_struct > >> + * instance and a copy of the value of module_region, just reuse the > >> + * allocation directly. > >> + */ > >> + work = module_region; > >> + INIT_WORK(work, module_free_wq_worker); > >> + queue_work(module_free_wq, work); > >> +} > > > > Whilst I can see that this works, it's a bit grotty because we're > > duplicating much of the hoop-jumping that vfree() also has to go through. > > Given that we allocate the module memory in the arch code, could we grab > > the pages at alloc time and stash them in the mod_arch_specific field? > > > > Yeah that should work Great! > >> +#endif > >> + > >> void *module_alloc(unsigned long size) > >> { > >> gfp_t gfp_mask = GFP_KERNEL; > >> @@ -65,6 +106,7 @@ void *module_alloc(unsigned long size) > >> return NULL; > >> } > >> > >> + remap_linear_module_alias(p, true); > >> return p; > >> } > >> > >> diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c > >> index 493ff75670ff..5492b691aafd 100644 > >> --- a/arch/arm64/mm/mmu.c > >> +++ b/arch/arm64/mm/mmu.c > >> @@ -432,7 +432,7 @@ static void __init map_mem(pgd_t *pgdp) > >> struct memblock_region *reg; > >> int flags = 0; > >> > >> - if (debug_pagealloc_enabled()) > >> + if (rodata_enabled || debug_pagealloc_enabled()) > >> flags = NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS; > > > > I suppose this is the most contentious part of the patch. Perhaps a > > better solution would be to tweak CONFIG_STRICT_MODULE_RWX so that it's > > structured more like CONFIG_STACKPROTECTOR. That way, your patch here > > could be the "strong" version, which comes with more of a performance > > cost. > > > > I'd like to understand whether this performance hit is theoretical or > not. It is obviously less efficient to map the linear region at page > granularity, but which workloads will actually notice the difference? > > Also, I think this should be a runtime setting, perhaps rodata=full? > With a CONFIG_ option to set the default. Yes, that's a better idea. We could even default to "full" if we don't have any performance data (if people start shouting then we'll know it's an issue!). Do you plan to spin a v2? Will
On 5 November 2018 at 20:31, Will Deacon <will.deacon@arm.com> wrote: > On Mon, Nov 05, 2018 at 01:29:34PM +0100, Ard Biesheuvel wrote: >> On 1 November 2018 at 16:17, Will Deacon <will.deacon@arm.com> wrote: >> > On Wed, Jul 11, 2018 at 07:04:26PM +0200, Ard Biesheuvel wrote: >> >> diff --git a/arch/arm64/include/asm/module.h b/arch/arm64/include/asm/module.h >> >> index 97d0ef12e2ff..f16f0073642b 100644 >> >> --- a/arch/arm64/include/asm/module.h >> >> +++ b/arch/arm64/include/asm/module.h >> >> @@ -91,4 +91,10 @@ static inline bool plt_entries_equal(const struct plt_entry *a, >> >> a->mov2 == b->mov2; >> >> } >> >> >> >> +#ifdef CONFIG_STRICT_KERNEL_RWX >> >> +void remap_linear_module_alias(void *module_region, bool ro); >> >> +#else >> >> +static inline void remap_linear_module_alias(void *module_region, bool ro) {} >> >> +#endif >> >> + >> >> #endif /* __ASM_MODULE_H */ >> >> diff --git a/arch/arm64/kernel/module.c b/arch/arm64/kernel/module.c >> >> index f0f27aeefb73..2b61ca0285eb 100644 >> >> --- a/arch/arm64/kernel/module.c >> >> +++ b/arch/arm64/kernel/module.c >> >> @@ -26,10 +26,51 @@ >> >> #include <linux/mm.h> >> >> #include <linux/moduleloader.h> >> >> #include <linux/vmalloc.h> >> >> +#include <linux/workqueue.h> >> >> #include <asm/alternative.h> >> >> #include <asm/insn.h> >> >> #include <asm/sections.h> >> >> >> >> +#ifdef CONFIG_STRICT_KERNEL_RWX >> >> + >> >> +static struct workqueue_struct *module_free_wq; >> >> + >> >> +static int init_workqueue(void) >> >> +{ >> >> + module_free_wq = alloc_ordered_workqueue("module_free_wq", 0); >> >> + WARN_ON(!module_free_wq); >> >> + >> >> + return 0; >> >> +} >> >> +pure_initcall(init_workqueue); >> >> + >> >> +static void module_free_wq_worker(struct work_struct *work) >> >> +{ >> >> + remap_linear_module_alias(work, false); >> >> + vfree(work); >> >> +} >> >> + >> >> +void module_memfree(void *module_region) >> >> +{ >> >> + struct work_struct *work; >> >> + >> >> + if (!module_region) >> >> + return; >> >> + >> >> + /* >> >> + * At this point, module_region is a pointer to an allocation of at >> >> + * least PAGE_SIZE bytes that is mapped read-write. So instead of >> >> + * allocating memory for a data structure containing a work_struct >> >> + * instance and a copy of the value of module_region, just reuse the >> >> + * allocation directly. >> >> + */ >> >> + work = module_region; >> >> + INIT_WORK(work, module_free_wq_worker); >> >> + queue_work(module_free_wq, work); >> >> +} >> > >> > Whilst I can see that this works, it's a bit grotty because we're >> > duplicating much of the hoop-jumping that vfree() also has to go through. >> > Given that we allocate the module memory in the arch code, could we grab >> > the pages at alloc time and stash them in the mod_arch_specific field? >> > >> >> Yeah that should work > > Great! > I may have spoken too soon: module_memfree() receives a void* pointer to the allocation to be freed, and we can't access the module struct from that. But looking into this a bit more, I wonder why this doesn not affect other architectures as well, and it might make sense to solve this generically, or at least, provide the hooks in generic code. >> >> +#endif >> >> + >> >> void *module_alloc(unsigned long size) >> >> { >> >> gfp_t gfp_mask = GFP_KERNEL; >> >> @@ -65,6 +106,7 @@ void *module_alloc(unsigned long size) >> >> return NULL; >> >> } >> >> >> >> + remap_linear_module_alias(p, true); >> >> return p; >> >> } >> >> >> >> diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c >> >> index 493ff75670ff..5492b691aafd 100644 >> >> --- a/arch/arm64/mm/mmu.c >> >> +++ b/arch/arm64/mm/mmu.c >> >> @@ -432,7 +432,7 @@ static void __init map_mem(pgd_t *pgdp) >> >> struct memblock_region *reg; >> >> int flags = 0; >> >> >> >> - if (debug_pagealloc_enabled()) >> >> + if (rodata_enabled || debug_pagealloc_enabled()) >> >> flags = NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS; >> > >> > I suppose this is the most contentious part of the patch. Perhaps a >> > better solution would be to tweak CONFIG_STRICT_MODULE_RWX so that it's >> > structured more like CONFIG_STACKPROTECTOR. That way, your patch here >> > could be the "strong" version, which comes with more of a performance >> > cost. >> > >> >> I'd like to understand whether this performance hit is theoretical or >> not. It is obviously less efficient to map the linear region at page >> granularity, but which workloads will actually notice the difference? >> >> Also, I think this should be a runtime setting, perhaps rodata=full? >> With a CONFIG_ option to set the default. > > Yes, that's a better idea. We could even default to "full" if we don't have > any performance data (if people start shouting then we'll know it's an > issue!). > > Do you plan to spin a v2? > Yes, but I'll do some digging first.
On 5 November 2018 at 21:18, Ard Biesheuvel <ard.biesheuvel@linaro.org> wrote: > On 5 November 2018 at 20:31, Will Deacon <will.deacon@arm.com> wrote: >> On Mon, Nov 05, 2018 at 01:29:34PM +0100, Ard Biesheuvel wrote: >>> On 1 November 2018 at 16:17, Will Deacon <will.deacon@arm.com> wrote: >>> > On Wed, Jul 11, 2018 at 07:04:26PM +0200, Ard Biesheuvel wrote: >>> >> diff --git a/arch/arm64/include/asm/module.h b/arch/arm64/include/asm/module.h >>> >> index 97d0ef12e2ff..f16f0073642b 100644 >>> >> --- a/arch/arm64/include/asm/module.h >>> >> +++ b/arch/arm64/include/asm/module.h >>> >> @@ -91,4 +91,10 @@ static inline bool plt_entries_equal(const struct plt_entry *a, >>> >> a->mov2 == b->mov2; >>> >> } >>> >> >>> >> +#ifdef CONFIG_STRICT_KERNEL_RWX >>> >> +void remap_linear_module_alias(void *module_region, bool ro); >>> >> +#else >>> >> +static inline void remap_linear_module_alias(void *module_region, bool ro) {} >>> >> +#endif >>> >> + >>> >> #endif /* __ASM_MODULE_H */ >>> >> diff --git a/arch/arm64/kernel/module.c b/arch/arm64/kernel/module.c >>> >> index f0f27aeefb73..2b61ca0285eb 100644 >>> >> --- a/arch/arm64/kernel/module.c >>> >> +++ b/arch/arm64/kernel/module.c >>> >> @@ -26,10 +26,51 @@ >>> >> #include <linux/mm.h> >>> >> #include <linux/moduleloader.h> >>> >> #include <linux/vmalloc.h> >>> >> +#include <linux/workqueue.h> >>> >> #include <asm/alternative.h> >>> >> #include <asm/insn.h> >>> >> #include <asm/sections.h> >>> >> >>> >> +#ifdef CONFIG_STRICT_KERNEL_RWX >>> >> + >>> >> +static struct workqueue_struct *module_free_wq; >>> >> + >>> >> +static int init_workqueue(void) >>> >> +{ >>> >> + module_free_wq = alloc_ordered_workqueue("module_free_wq", 0); >>> >> + WARN_ON(!module_free_wq); >>> >> + >>> >> + return 0; >>> >> +} >>> >> +pure_initcall(init_workqueue); >>> >> + >>> >> +static void module_free_wq_worker(struct work_struct *work) >>> >> +{ >>> >> + remap_linear_module_alias(work, false); >>> >> + vfree(work); >>> >> +} >>> >> + >>> >> +void module_memfree(void *module_region) >>> >> +{ >>> >> + struct work_struct *work; >>> >> + >>> >> + if (!module_region) >>> >> + return; >>> >> + >>> >> + /* >>> >> + * At this point, module_region is a pointer to an allocation of at >>> >> + * least PAGE_SIZE bytes that is mapped read-write. So instead of >>> >> + * allocating memory for a data structure containing a work_struct >>> >> + * instance and a copy of the value of module_region, just reuse the >>> >> + * allocation directly. >>> >> + */ >>> >> + work = module_region; >>> >> + INIT_WORK(work, module_free_wq_worker); >>> >> + queue_work(module_free_wq, work); >>> >> +} >>> > >>> > Whilst I can see that this works, it's a bit grotty because we're >>> > duplicating much of the hoop-jumping that vfree() also has to go through. >>> > Given that we allocate the module memory in the arch code, could we grab >>> > the pages at alloc time and stash them in the mod_arch_specific field? >>> > >>> >>> Yeah that should work >> >> Great! >> > > I may have spoken too soon: module_memfree() receives a void* pointer > to the allocation to be freed, and we can't access the module struct > from that. > > But looking into this a bit more, I wonder why this doesn not affect > other architectures as well, and it might make sense to solve this > generically, or at least, provide the hooks in generic code. > >>> >> +#endif >>> >> + >>> >> void *module_alloc(unsigned long size) >>> >> { >>> >> gfp_t gfp_mask = GFP_KERNEL; >>> >> @@ -65,6 +106,7 @@ void *module_alloc(unsigned long size) >>> >> return NULL; >>> >> } >>> >> >>> >> + remap_linear_module_alias(p, true); >>> >> return p; >>> >> } >>> >> >>> >> diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c >>> >> index 493ff75670ff..5492b691aafd 100644 >>> >> --- a/arch/arm64/mm/mmu.c >>> >> +++ b/arch/arm64/mm/mmu.c >>> >> @@ -432,7 +432,7 @@ static void __init map_mem(pgd_t *pgdp) >>> >> struct memblock_region *reg; >>> >> int flags = 0; >>> >> >>> >> - if (debug_pagealloc_enabled()) >>> >> + if (rodata_enabled || debug_pagealloc_enabled()) >>> >> flags = NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS; >>> > >>> > I suppose this is the most contentious part of the patch. Perhaps a >>> > better solution would be to tweak CONFIG_STRICT_MODULE_RWX so that it's >>> > structured more like CONFIG_STACKPROTECTOR. That way, your patch here >>> > could be the "strong" version, which comes with more of a performance >>> > cost. >>> > >>> >>> I'd like to understand whether this performance hit is theoretical or >>> not. It is obviously less efficient to map the linear region at page >>> granularity, but which workloads will actually notice the difference? >>> >>> Also, I think this should be a runtime setting, perhaps rodata=full? >>> With a CONFIG_ option to set the default. >> >> Yes, that's a better idea. We could even default to "full" if we don't have >> any performance data (if people start shouting then we'll know it's an >> issue!). >> >> Do you plan to spin a v2? >> > > Yes, but I'll do some digging first. OK, so the answer is that other architectures (or at least, x86) does this at a lower level. The set_memory_{ro,rw,x,nx} take care of the permission attributes of the linear alias any time they are modified. So for arm64, that is probably the way to go as well. I'll update my patches accordingly.
diff --git a/arch/arm64/include/asm/module.h b/arch/arm64/include/asm/module.h index 97d0ef12e2ff..f16f0073642b 100644 --- a/arch/arm64/include/asm/module.h +++ b/arch/arm64/include/asm/module.h @@ -91,4 +91,10 @@ static inline bool plt_entries_equal(const struct plt_entry *a, a->mov2 == b->mov2; } +#ifdef CONFIG_STRICT_KERNEL_RWX +void remap_linear_module_alias(void *module_region, bool ro); +#else +static inline void remap_linear_module_alias(void *module_region, bool ro) {} +#endif + #endif /* __ASM_MODULE_H */ diff --git a/arch/arm64/kernel/module.c b/arch/arm64/kernel/module.c index f0f27aeefb73..2b61ca0285eb 100644 --- a/arch/arm64/kernel/module.c +++ b/arch/arm64/kernel/module.c @@ -26,10 +26,51 @@ #include <linux/mm.h> #include <linux/moduleloader.h> #include <linux/vmalloc.h> +#include <linux/workqueue.h> #include <asm/alternative.h> #include <asm/insn.h> #include <asm/sections.h> +#ifdef CONFIG_STRICT_KERNEL_RWX + +static struct workqueue_struct *module_free_wq; + +static int init_workqueue(void) +{ + module_free_wq = alloc_ordered_workqueue("module_free_wq", 0); + WARN_ON(!module_free_wq); + + return 0; +} +pure_initcall(init_workqueue); + +static void module_free_wq_worker(struct work_struct *work) +{ + remap_linear_module_alias(work, false); + vfree(work); +} + +void module_memfree(void *module_region) +{ + struct work_struct *work; + + if (!module_region) + return; + + /* + * At this point, module_region is a pointer to an allocation of at + * least PAGE_SIZE bytes that is mapped read-write. So instead of + * allocating memory for a data structure containing a work_struct + * instance and a copy of the value of module_region, just reuse the + * allocation directly. + */ + work = module_region; + INIT_WORK(work, module_free_wq_worker); + queue_work(module_free_wq, work); +} + +#endif + void *module_alloc(unsigned long size) { gfp_t gfp_mask = GFP_KERNEL; @@ -65,6 +106,7 @@ void *module_alloc(unsigned long size) return NULL; } + remap_linear_module_alias(p, true); return p; } diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c index 493ff75670ff..5492b691aafd 100644 --- a/arch/arm64/mm/mmu.c +++ b/arch/arm64/mm/mmu.c @@ -432,7 +432,7 @@ static void __init map_mem(pgd_t *pgdp) struct memblock_region *reg; int flags = 0; - if (debug_pagealloc_enabled()) + if (rodata_enabled || debug_pagealloc_enabled()) flags = NO_BLOCK_MAPPINGS | NO_CONT_MAPPINGS; /* diff --git a/arch/arm64/mm/pageattr.c b/arch/arm64/mm/pageattr.c index a56359373d8b..cc04be572660 100644 --- a/arch/arm64/mm/pageattr.c +++ b/arch/arm64/mm/pageattr.c @@ -138,6 +138,25 @@ int set_memory_valid(unsigned long addr, int numpages, int enable) __pgprot(PTE_VALID)); } +#ifdef CONFIG_STRICT_KERNEL_RWX +void remap_linear_module_alias(void *module_region, bool ro) +{ + struct vm_struct *vm; + int i; + + if (!rodata_enabled) + return; + + vm = find_vm_area(module_region); + WARN_ON(!vm || !vm->pages); + + for (i = 0; i < vm->nr_pages; i++) + __change_memory_common((u64)page_address(vm->pages[i]), PAGE_SIZE, + ro ? __pgprot(PTE_RDONLY) : __pgprot(PTE_WRITE), + ro ? __pgprot(PTE_WRITE) : __pgprot(PTE_RDONLY)); +} +#endif + #ifdef CONFIG_DEBUG_PAGEALLOC void __kernel_map_pages(struct page *page, int numpages, int enable) {
When CONFIG_STRICT_KERNEL_RWX=y [which is the default on arm64], we take great care to ensure that the mappings of modules in the vmalloc space are locked down as much as possible, i.e., executable code is mapped read-only, read-only data is mapped read-only and non-executable, and read-write data is mapped non-executable as well. However, due to the way we map the linear region [aka the kernel direct mapping], those module regions are aliased by read-write mappings, and it is possible for an attacker to modify code or data that was assumed to be immutable in this configuration. So let's ensure that the linear alias of module memory is remapped read-only upon allocation and remapped read-write when it is freed. The latter requires some special handling involving a workqueue due to the fact that it may be called in softirq context at which time calling find_vm_area() is unsafe. Note that this requires the entire linear region to be mapped down to pages, which may result in a performance regression in some hardware configurations. Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> --- Changes since RFC/v1: - remap linear alias read-only rather than invalid - honour rodata_enabled, i.e., 'rodata=off' will disable this functionality - use VMA nr_pages field rather than size to obtain the number of pages to remap arch/arm64/include/asm/module.h | 6 +++ arch/arm64/kernel/module.c | 42 ++++++++++++++++++++ arch/arm64/mm/mmu.c | 2 +- arch/arm64/mm/pageattr.c | 19 +++++++++ 4 files changed, 68 insertions(+), 1 deletion(-)