mm: Unexport apply_to_page_range()
Subsystems (7)
cover mm: Unexport apply_to_page_range()
Hi,
These patches remove the last few modular apply_to_page_range() users and unexports these interface. Since these functions provide direct access to our page-tables they're a prime target for nefarious activities.
These patches rely on the remap_pfn_range_notrack() patches from Christoph that are currently already in Andrew's tree.
Please consider.
PATCH 1/7 mm: Unexport apply_to_existing_page_range()
There are no modular in-tree users, remove the EXPORT.
This is an unsafe function in that it gives direct access to the page-tables.
--- mm/memory.c | 1 - 1 file changed, 1 deletion(-)
mm/memory.c +0 -1
@@ -2558,7 +2558,6 @@ int apply_to_existing_page_range(struct {return __apply_to_page_range (mm, addr, size, fn, data, false); } -EXPORT_SYMBOL_GPL (apply_to_existing_page_range);/* * handle_pte_fault chooses page fault handler according to an entry which was
Re: [PATCH 1/7] mm: Unexport apply_to_existing_page_range()
Quoted 6 lines · Peter Zijlstra
There are no modular in-tree users, remove the EXPORT.
This is an unsafe function in that it gives direct access to the page-tables.
Signed-off-by Peter Zijlstra (Intel) <[email protected]>
Looks good,
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 2/7 xen/gntdev,x86: Remove apply_to_page_range() use from module
In reply to Peter ZijlstraInstead of relying on apply_to_page_range() being available to modules, move its use into core kernel code and export it's application.
NOTE: ideally we do: use_ptemod = !auto_translate_physmap && gnttab_map_avail_bits and remove this hack.
--- arch/x86/include/asm/xen/page.h | 2 ++ arch/x86/xen/mmu.c | 26 ++++++++++++++++++++++++++ drivers/xen/gntdev.c | 23 +---------------------- 3 files changed, 29 insertions(+), 22 deletions(-)
arch/x86/include/asm/xen/page.h +2 -0
@@ -370,4 +370,6 @@ static inline unsigned long xen_get_swio return __get_free_pages(__GFP_NOWARN, order); } +extern void xen_set_grant_as_special (struct vm_area_struct *vma); +#endif /* _ASM_X86_XEN_PAGE_H */
arch/x86/xen/mmu.c +26 -0
@@ -51,3 +51,29 @@ int xen_unmap_domain_gfn_range(struct vmreturn -EINVAL ; }EXPORT_SYMBOL_GPL (xen_unmap_domain_gfn_range); + +static int set_grant_ptes_as_special (pte_t *pte,unsigned long addr,void *data) +{ +set_pte_at (current->mm, addr, pte,pte_mkspecial (*pte)); +return 0 ; +} + +void xen_set_grant_as_special (struct vm_area_struct *vma) +{ +if (xen_feature (XENFEAT_gnttab_map_avail_bits)) +return ; + +/* +* If the PTEs were not made special by the grant map +* hypercall, do so here. +* +* This is racy since the mapping is already visible +* to userspace but userspace should be well-behaved +* enough to not touch it until the mmap() call +* returns. +*/ +apply_to_page_range (vma->vm_mm, vma->vm_start, + vma->vm_end - vma->vm_start, + set_grant_ptes_as_special,NULL ); +} +EXPORT_SYMBOL_GPL (xen_set_grant_as_special);
drivers/xen/gntdev.c +1 -22
@@ -278,14 +278,6 @@ static int find_grant_ptes(pte_t *pte, ureturn 0 ; } -#ifdef CONFIG_X86 -static int set_grant_ptes_as_special (pte_t *pte,unsigned long addr,void *data) -{ -set_pte_at (current->mm, addr, pte,pte_mkspecial (*pte)); -return 0 ; -} -#endif -int gntdev_map_grant_pages (struct gntdev_grant_map *map) {int i, err =0 ;
@@ -1040,20 +1032,7 @@ static int gntdev_mmap(struct file *flip goto out_put_map; }else {#ifdef CONFIG_X86 -/* -* If the PTEs were not made special by the grant map -* hypercall, do so here. -* -* This is racy since the mapping is already visible -* to userspace but userspace should be well-behaved -* enough to not touch it until the mmap() call -* returns. -*/ -if (!xen_feature (XENFEAT_gnttab_map_avail_bits)) { -apply_to_page_range (vma->vm_mm, vma->vm_start, - vma->vm_end - vma->vm_start, - set_grant_ptes_as_special,NULL ); - } +xen_set_grant_as_special (vma);#endif }
Re: [PATCH 2/7] xen/gntdev,x86: Remove apply_to_page_range() use from module
Quoted 3 lines · Peter Zijlstra
Instead of relying on
apply_to_page_range()being available to modules, move its use into core kernel code and export it's application.
This doesn't exactly look great, but at least it contains the damage..
Quoted 3 lines
NOTE: ideally we do:
use_ptemod= !auto_translate_physmap &&gnttab_map_avail_bitsand remove this hack.
Given how much pain the !auto_translate_physmap case causes all over the kernel I wonder what a realistic timeline might be for dropping support for this case might be..
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 2/7] xen/gntdev,x86: Remove apply_to_page_range() use from module
Quoted 6 lines · Christoph Hellwig
Quoted 3 lines · Peter Zijlstra
Instead of relying on
apply_to_page_range()being available to modules, move its use into core kernel code and export it's application.This doesn't exactly look great, but at least it contains the damage..
That was just about as far as I got...
Quoted 6 lines
Quoted 2 lines
NOTE: ideally we do:
use_ptemod= !auto_translate_physmap &&gnttab_map_avail_bitsand remove this hack.Given how much pain the !auto_translate_physmap case causes all over the kernel I wonder what a realistic timeline might be for dropping support for this case might be..
I've no experience with any of that; it looks absolutely disguisting, all of it. I figured that's part of the Xen 'charm' :-)
Juergen Gross
maintainerRe: [PATCH 2/7] xen/gntdev,x86: Remove apply_to_page_range() use from module
In reply to Christoph HellwigQuoted 8 lines · Christoph Hellwig
Quoted 3 lines · Peter Zijlstra
Instead of relying on
apply_to_page_range()being available to modules, move its use into core kernel code and export it's application.This doesn't exactly look great, but at least it contains the damage..
Quoted 1 line
Quoted 6 lines
Quoted 2 lines
NOTE: ideally we do:
use_ptemod= !auto_translate_physmap &&gnttab_map_avail_bitsand remove this hack.Given how much pain the !auto_translate_physmap case causes all over the kernel I wonder what a realistic timeline might be for dropping support for this case might be..
Think in the order of years.
It is basically the Xen PV guest support you are speaking of here, and the planned replacement PVH especially for dom0 is still lacking some functionality and it has performance issues.
Juergen
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 3/7 xen/gntdev: Remove apply_to_page_range() use from module
In reply to Peter ZijlstraInstead of relying on apply_to_page_range() being available to modules, move its use into core kernel code and export it's application.
--- drivers/xen/gntdev-common.h | 2 ++ drivers/xen/gntdev.c | 30 +----------------------------- drivers/xen/grant-table.c | 37 +++++++++++++++++++++++++++++++++++++ 3 files changed, 40 insertions(+), 29 deletions(-)
drivers/xen/gntdev-common.h +2 -0
@@ -86,4 +86,6 @@ bool gntdev_test_page_count(unsigned intint gntdev_map_grant_pages (struct gntdev_grant_map *map); +int gnttab_use_ptemod (struct vm_area_struct *vma,struct gntdev_grant_map *map); +#endif
drivers/xen/gntdev.c +1 -29
@@ -262,32 +262,6 @@ void gntdev_put_map(struct gntdev_priv */* ------------------------------------------------------------------ */ -static int find_grant_ptes (pte_t *pte,unsigned long addr,void *data) -{ -struct gntdev_grant_map *map = data; -unsigned int pgnr = (addr - map->vma->vm_start) >>PAGE_SHIFT ; -int flags = map->flags | GNTMAP_application_map | GNTMAP_contains_pte; -u64 pte_maddr; - -BUG_ON (pgnr >= map->count); - pte_maddr =arbitrary_virt_to_machine (pte).maddr; - -/* -* Set the PTE as special to force get_user_pages_fast() fall -* back to the slow path. If this is not supported as part of -* the grant map, it will be done afterwards. -*/ -if (xen_feature (XENFEAT_gnttab_map_avail_bits)) - flags |= (1 << _GNTMAP_guest_avail0); - -gnttab_set_map_op (&map->map_ops[pgnr], pte_maddr, flags, - map->grants[pgnr].ref, - map->grants[pgnr].domid); -gnttab_set_unmap_op (&map->unmap_ops[pgnr], pte_maddr, flags, -INVALID_GRANT_HANDLE ); -return 0 ; -} -int gntdev_map_grant_pages (struct gntdev_grant_map *map) {int i, err =0 ;
@@ -1028,9 +1002,7 @@ static int gntdev_mmap(struct file *flipmmu_interval_read_begin (&map->notifier); map->pages_vm_start = vma->vm_start; - err =apply_to_page_range (vma->vm_mm, vma->vm_start, - vma->vm_end - vma->vm_start, - find_grant_ptes, map); + err =gnttab_use_ptemod (vma, map);if (err) {pr_warn ("find_grant_ptes() failure.\n" ); goto out_put_map;
drivers/xen/grant-table.c +37 -0
@@ -1591,6 +1591,43 @@ int gnttab_init(void) }EXPORT_SYMBOL_GPL (gnttab_init); +#include <xen/gntdev.h> +#include "gntdev-common.h" + +static int find_grant_ptes (pte_t *pte,unsigned long addr,void *data) +{ +struct gntdev_grant_map *map = data; +unsigned int pgnr = (addr - map->vma->vm_start) >>PAGE_SHIFT ; +int flags = map->flags | GNTMAP_application_map | GNTMAP_contains_pte; +u64 pte_maddr; + +BUG_ON (pgnr >= map->count); + pte_maddr =arbitrary_virt_to_machine (pte).maddr; + +/* +* Set the PTE as special to force get_user_pages_fast() fall +* back to the slow path. If this is not supported as part of +* the grant map, it will be done afterwards. +*/ +if (xen_feature (XENFEAT_gnttab_map_avail_bits)) + flags |= (1 << _GNTMAP_guest_avail0); + +gnttab_set_map_op (&map->map_ops[pgnr], pte_maddr, flags, + map->grants[pgnr].ref, + map->grants[pgnr].domid); +gnttab_set_unmap_op (&map->unmap_ops[pgnr], pte_maddr, flags, +INVALID_GRANT_HANDLE ); +return 0 ; +} + +int gnttab_use_ptemod (struct vm_area_struct *vma,struct gntdev_grant_map *map) +{ +return apply_to_page_range (vma->vm_mm, vma->vm_start, + vma->vm_end - vma->vm_start, + find_grant_ptes, map); +} +EXPORT_SYMBOL_GPL (gnttab_use_ptemod); +static int __gnttab_init (void ) {if (!xen_domain ())
Re: [PATCH 3/7] xen/gntdev: Remove apply_to_page_range() use from module
drivers/xen/grant-table.c int gnttab_init(void) · line 1595
Quoted 8 lines · Peter Zijlstra
drivers/xen/grant-table.c +2
-0@@ -1591,6 +1591,43 @@ int gnttab_init(void) }EXPORT_SYMBOL_GPL (gnttab_init); +#include <xen/gntdev.h> +#include "gntdev-common.h"
Can't we keep the includes at the top of the file?
Otherwise looks good:
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 3/7] xen/gntdev: Remove apply_to_page_range() use from module
Quoted 11 lines · Christoph Hellwig
drivers/xen/grant-table.c int gnttab_init(void) · line 1595
Quoted 8 lines · Peter Zijlstra
drivers/xen/grant-table.c +2
-0@@ -1591,6 +1591,43 @@ int gnttab_init(void) }EXPORT_SYMBOL_GPL (gnttab_init); +#include <xen/gntdev.h> +#include "gntdev-common.h" Can't we keep the includes at the top of the file?
Probably could, lemme move them.
PATCH 4/7 mm: Introduce verify_page_range()
In reply to Peter ZijlstraIntroduce and EXPORT a read-only counterpart to apply_to_page_range().
It only exposes the PTE value, not a pointer to the pagetables itself and is thus quite a bit safer to export. A number of apply_to_page_range() users can be converted to this primitive.
--- include/linux/mm.h | 4 ++++ mm/memory.c | 24 ++++++++++++++++++++++++ 2 files changed, 28 insertions(+)
include/linux/mm.h +4 -0
@@ -2876,6 +2876,10 @@ extern int apply_to_page_range(struct mmextern int apply_to_existing_page_range (struct mm_struct *mm,unsigned long address,unsigned long size,pte_fn_t fn,void *data); +extern int verify_page_range (struct mm_struct *mm, +unsigned long addr,unsigned long size, +int (*fn)(pte_t pte,unsigned long addr,void *data), +void *data);extern void init_mem_debugging_and_hardening (void );#ifdef CONFIG_PAGE_POISONING
mm/memory.c +24 -0
@@ -2559,6 +2559,30 @@ int apply_to_existing_page_range(structreturn __apply_to_page_range (mm, addr, size, fn, data, false); } +struct vpr_data { +int (*fn)(pte_t pte,unsigned long addr,void *data); +void *data; +}; + +static int vpr_fn (pte_t *pte,unsigned long addr,void *data) +{ +struct vpr_data *vpr = data; +return vpr->fn (*pte, addr, vpr->data); +} + +int verify_page_range (struct mm_struct *mm, +unsigned long addr,unsigned long size, +int (*fn)(pte_t pte,unsigned long addr,void *data), +void *data) +{ +struct vpr_data vpr = { + .fn = fn, + .data = data, + }; +return apply_to_page_range (mm, addr, size, vpr_fn, &vpr); +} +EXPORT_SYMBOL_GPL (verify_page_range); +/* * handle_pte_fault chooses page fault handler according to an entry which was * read non-atomically. Before making any commitment, on those architectures
Re: [PATCH 4/7] mm: Introduce verify_page_range()
Quoted 1 line · Peter Zijlstra
+extern int verify_page_range(struct
mm_struct*mm,
No need for the extern here.
mm/memory.c int apply_to_existing_page_range(struct · line 2576
Quoted 4 lines
mm/memory.c +4
-0+int verify_page_range (struct mm_struct *mm, +unsigned long addr,unsigned long size, +int (*fn)(pte_t pte,unsigned long addr,void *data), +void *data)
A kerneldoc comment would be nice for this function.
Otherwise this looks fine.
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 4/7] mm: Introduce verify_page_range()
Quoted 4 lines · Christoph Hellwig
Quoted 1 line · Peter Zijlstra
+extern int verify_page_range(struct
mm_struct*mm,No need for the extern here.
It's consistent with the rest of the functions there. Also, I personally like that extern.
Quoted 8 lines
mm/memory.c int apply_to_existing_page_range(struct · line 2576
Quoted 4 lines
mm/memory.c +4
-0+int verify_page_range (struct mm_struct *mm, +unsigned long addr,unsigned long size, +int (*fn)(pte_t pte,unsigned long addr,void *data), +void *data)A kerneldoc comment would be nice for this function.
Otherwise this looks fine.
Something like so?
/**
* verify_page_range() - Scan (and fill) a range of virtual memory and validate PTEs
* @mm: mm identifying the virtual memory map
* @addr: starting virtual address of the range
* @size: size of the range
* @fn: function that verifies the PTEs
* @data: opaque data passed to @fn
*
* Scan a region of virtual memory, filling in page tables as necessary and
* calling a provided function on each leaf, providing a copy of the
* page-table-entry.
*
* Similar apply_to_page_range(), but does not provide direct access to the
* page-tables.
*
* NOTE! this function does not work correctly vs large pages.
*
* Return: the first !0 return of the provided function, or 0 on completion. */
int verify_page_range(struct mm_struct *mm,
unsigned long addr, unsigned long size,
int (*fn)(pte_t pte, unsigned long addr, void *data), void *data)
Re: [PATCH 4/7] mm: Introduce verify_page_range()
In reply to Peter Zijlstramm/memory.c int apply_to_existing_page_range(struct · line 2565
Quoted 4 lines · Peter Zijlstra
mm/memory.c +4
-0+struct vpr_data { +int (*fn)(pte_t pte,unsigned long addr,void *data); +void *data; +};
Eeerg. This is likely to become an attack target itself. Stored function pointer with stored (3rd) argument.
This doesn't seem needed: only DRM uses it, and that's for error reporting. I'd rather plumb back errors in a way to not have to add another place in the kernel where we do func+arg stored calling.
Signature · 5 lines
Kees Cook _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 4/7] mm: Introduce verify_page_range()
Quoted 8 lines · Kees Cook
mm/memory.c int apply_to_existing_page_range(struct · line 2565
Quoted 4 lines · Peter Zijlstra
mm/memory.c +4
-0+struct vpr_data { +int (*fn)(pte_t pte,unsigned long addr,void *data); +void *data; +};Eeerg. This is likely to become an attack target itself. Stored function pointer with stored (3rd) argument.
You got some further reading on that? How exactly are those exploited?
Re: [PATCH 4/7] mm: Introduce verify_page_range()
Quoted 11 lines · Peter Zijlstra
Quoted 8 lines · Kees Cook
mm/memory.c int apply_to_existing_page_range(struct · line 2565
Quoted 4 lines · Peter Zijlstra
mm/memory.c +4
-0+struct vpr_data { +int (*fn)(pte_t pte,unsigned long addr,void *data); +void *data; +};Eeerg. This is likely to become an attack target itself. Stored function pointer with stored (3rd) argument.
You got some further reading on that? How exactly are those exploited?
Sure, see "Executing code" in
https://googleprojectzero.blogspot.com/2017/05/exploiting-linux-kernel-via-packet.html
I killed the entire primitive (for timer_list)
https://outflux.net/blog/archives/2018/02/05/security-things-in-linux-v4-15/#v4.15-timer_list
but that was a lot of work, so I'm trying to avoid seeing more things like it appear. :) (And I'm trying to get rid of similar APIs, like tasklet.)
This new code is unlikely to ever be used as widely as timer_list, but I just cringe when I see the code pattern. I'll understand if there isn't a solution that doesn't require major refactoring, but I can dream. :)
Signature · 5 lines
Kees Cook _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 4/7] mm: Introduce verify_page_range()
In reply to Kees CookQuoted 9 lines · Kees Cook
mm/memory.c int apply_to_existing_page_range(struct · line 2565
Quoted 4 lines · Peter Zijlstra
mm/memory.c +4
-0+struct vpr_data { +int (*fn)(pte_t pte,unsigned long addr,void *data); +void *data; +};Eeerg. This is likely to become an attack target itself. Stored function pointer with stored (3rd) argument.
Quoted 3 lines
This doesn't seem needed: only DRM uses it, and that's for error reporting. I'd rather plumb back errors in a way to not have to add another place in the kernel where we do func+arg stored calling.
Is this any better? It does have the stored pointer, but not a stored argument, assuming you don't count returns as arguments I suppose.
The alternative is refactoring apply_to_page_range() :-/
---
struct vpr_data {
bool (*fn)(pte_t pte, unsigned long addr);
unsigned long addr;
};static int vpr_fn(pte_t *pte, unsigned long addr, void *data)
{
struct vpr_data *vpr = data;
if (!vpr->fn(*pte, addr)) {
vpr->addr = addr;
return -EINVAL;
}
return 0;
}/**
* verify_page_range() - Scan (and fill) a range of virtual memory and validate PTEs
* @mm: mm identifying the virtual memory map
* @addr: starting virtual address of the range
* @size: size of the range
* @fn: function that verifies the PTEs
*
* Scan a region of virtual memory, filling in page tables as necessary and
* calling a provided function on each leaf, providing a copy of the
* page-table-entry.
*
* Similar apply_to_page_range(), but does not provide direct access to the
* page-tables.
*
* NOTE! this function does not work correctly vs large pages.
*
* Return: the address that failed verification or 0 on success.
*/
unsigned long verify_page_range(struct mm_struct *mm,
unsigned long addr, unsigned long size, bool (*fn)(pte_t pte, unsigned long addr))
{
struct vpr_data vpr = {
.fn = fn,
.addr = 0,
};
apply_to_page_range(mm, addr, size, vpr_fn, &vpr);
return vpr.addr;
}EXPORT_SYMBOL_GPL(verify_page_range);
Re: [PATCH 4/7] mm: Introduce verify_page_range()
Quoted 10 lines · Peter Zijlstra
Quoted 9 lines · Kees Cook
mm/memory.c int apply_to_existing_page_range(struct · line 2565
Quoted 4 lines · Peter Zijlstra
mm/memory.c +4
-0+struct vpr_data { +int (*fn)(pte_t pte,unsigned long addr,void *data); +void *data; +};Eeerg. This is likely to become an attack target itself. Stored function pointer with stored (3rd) argument.
Quoted 6 lines
Quoted 3 lines
This doesn't seem needed: only DRM uses it, and that's for error reporting. I'd rather plumb back errors in a way to not have to add another place in the kernel where we do func+arg stored calling.
Is this any better? It does have the stored pointer, but not a stored argument, assuming you don't count returns as arguments I suppose.
It's better in the sense that it's not the func/arg pair that really bugs me, yes. :)
Quoted 1 line
The alternative is refactoring
apply_to_page_range():-/
Yeah, I'm looking now, I see what you mean.
Quoted 8 lines
---
struct vpr_data { bool (*fn)(pte_t pte, unsigned long addr); unsigned long addr; };static int
vpr_fn(pte_t *pte, unsigned long addr, void *data)
Quoted 8 lines
{
struct vpr_data *vpr = data;
if (!vpr->fn(*pte, addr)) {
vpr->addr = addr;
return -EINVAL;
}
return 0;
}My point about passing "addr" was that nothing in the callback actually needs it -- the top level can just as easily report the error. And that the helper is always vpr_fn(), so it doesn't need to be passed either.
So the addr can just be encoded in "int", and no structure is needed at:
typedef bool (*vpr_fn_t)(pte_t pte);
static int vpr_fn(pte_t *pte, unsigned long addr, void *data)
{
vpr_fn_t callback = data;
if (!callback(*pte))
return addr >> PAGE_SIZE;
return 0;
}unsigned long verify_page_range(struct mm_struct *mm,
unsigned long addr, unsigned long size, vpr_fn_t callback)
{
return apply_to_page_range(mm, addr, size, vpr_fn, callback) << PAGE_SIZE;
}But maybe I'm missing something?
Signature · 5 lines
Kees Cook _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 4/7] mm: Introduce verify_page_range()
Quoted 13 lines · Kees Cook
So the addr can just be encoded in "int", and no structure is needed at:
typedef bool (*vpr_fn_t)(pte_t pte);
static int
vpr_fn(pte_t *pte, unsigned long addr, void *data){ vpr_fn_t callback = data; if (!callback(*pte)) return addr >> PAGE_SIZE; return 0; }
Quoted 8 lines
unsigned long verify_page_range(struct
mm_struct*mm,
unsigned long addr, unsigned long size,vpr_fn_tcallback){ return apply_to_page_range(mm, addr, size, vpr_fn, callback) << PAGE_SIZE; }But maybe I'm missing something?
That covers only (32+12) bits of address space and will mostly work, but we definitely support architectures (very much including x86_64) with larger address spaces than that.
PATCH 5/7 xen/privcmd: Use verify_page_range()
In reply to Peter ZijlstraAvoid using apply_to_page_range() from modules, use the safer verify_page_range() instead.
--- drivers/xen/privcmd.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
drivers/xen/privcmd.c +4 -4
@@ -946,9 +946,9 @@ static int privcmd_mmap(struct file *fil* on a per pfn/pte basis. Mapping calls that fail with ENOENT * can be then retried until success. */ -static int is_mapped_fn (pte_t *pte,unsigned long addr,void *data) +static int is_mapped_fn (pte_t pte,unsigned long addr,void *data) { -return pte_none (*pte) ?0 : -EBUSY ; +return pte_none (pte) ?0 : -EBUSY ; } static int privcmd_vma_range_is_mapped(
@@ -956,8 +956,8 @@ static int privcmd_vma_range_is_mapped( unsigned long addr, unsigned long nr_pages) { -return apply_to_page_range (vma->vm_mm, addr, nr_pages <<PAGE_SHIFT , - is_mapped_fn,NULL ) !=0 ; +return verify_page_range (vma->vm_mm, addr, nr_pages <<PAGE_SHIFT , + is_mapped_fn,NULL ) !=0 ; }const struct file_operations xen_privcmd_fops = {
Re: [PATCH 5/7] xen/privcmd: Use verify_page_range()
Looks good,
PATCH 6/7 i915: Convert to verify_page_range()
In reply to Peter Zijlstracheck_{present,absent}() only need R/O access, use verify_page_range() instead to remove modular use of apply_to_page_range().
--- drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-)
drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c +8 -8
@@ -1225,9 +1225,9 @@ static int igt_mmap_gpu(void *arg)return 0 ; } -static int check_present_pte (pte_t *pte,unsigned long addr,void *data) +static int check_present_pte (pte_t pte,unsigned long addr,void *data) { -if (!pte_present (*pte) ||pte_none (*pte)) { +if (!pte_present (pte) ||pte_none (pte)) {pr_err ("missing PTE:%lx\n" , (addr - (unsigned long )data) >>PAGE_SHIFT );return -EINVAL ;
@@ -1236,9 +1236,9 @@ static int check_present_pte(pte_t *pte,return 0 ; } -static int check_absent_pte (pte_t *pte,unsigned long addr,void *data) +static int check_absent_pte (pte_t pte,unsigned long addr,void *data) { -if (pte_present (*pte) && !pte_none (*pte)) { +if (pte_present (pte) && !pte_none (pte)) {pr_err ("present PTE:%lx; expected to be revoked\n" , (addr - (unsigned long )data) >>PAGE_SHIFT );return -EINVAL ;
@@ -1249,14 +1249,14 @@ static int check_absent_pte(pte_t *pte,static int check_present (unsigned long addr,unsigned long len) { -return apply_to_page_range (current->mm, addr, len, - check_present_pte, (void *)addr); +return verify_page_range (current->mm, addr, len, + check_present_pte, (void *)addr); }static int check_absent (unsigned long addr,unsigned long len) { -return apply_to_page_range (current->mm, addr, len, - check_absent_pte, (void *)addr); +return verify_page_range (current->mm, addr, len, + check_absent_pte, (void *)addr); }static int prefault_range (u64 start,u64 len)
Re: [PATCH 6/7] i915: Convert to verify_page_range()
Looks good,
Re: [PATCH 6/7] i915: Convert to verify_page_range()
In reply to Peter ZijlstraQuoted 35 lines · Peter Zijlstra
check_{present,absent}() only need R/O access, use
verify_page_range()instead to remove modular use ofapply_to_page_range().Signed-off-by Peter Zijlstra (Intel) <[email protected]>--- drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-)drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c +4
-4@@ -1225,9 +1225,9 @@ static int igt_mmap_gpu(void *arg)return 0 ; } -static int check_present_pte (pte_t *pte,unsigned long addr,void *data) +static int check_present_pte (pte_t pte,unsigned long addr,void *data) { -if (!pte_present (*pte) ||pte_none (*pte)) { +if (!pte_present (pte) ||pte_none (pte)) {pr_err ("missing PTE:%lx\n" , (addr - (unsigned long )data) >>PAGE_SHIFT );return -EINVAL ;@@ -1236,9 +1236,9 @@ static int check_present_pte(pte_t *pte,return 0 ; } -static int check_absent_pte (pte_t *pte,unsigned long addr,void *data) +static int check_absent_pte (pte_t pte,unsigned long addr,void *data) { -if (pte_present (*pte) && !pte_none (*pte)) { +if (pte_present (pte) && !pte_none (pte)) {pr_err ("present PTE:%lx; expected to be revoked\n" , (addr - (unsigned long )data) >>PAGE_SHIFT );return -EINVAL ;
drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c static int check_absent_pte(pte_t *pte, · line 1253
Quoted 7 lines
drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c +2
-2@@ -1249,14 +1249,14 @@ static int check_absent_pte(pte_t *pte,static int check_present (unsigned long addr,unsigned long len) { -return apply_to_page_range (current->mm, addr, len, - check_present_pte, (void *)addr); +return verify_page_range (current->mm, addr, len, + check_present_pte, (void *)addr);
For example, switch to returning bad addr through verify_page_range(), or have a by-reference value, etc:
unsigned long failed;
failed = verify_page_range(current->mm< addr, len, check_present_pte);
if (failed) { pr_err("missing PTE:%lx\n",
(addr - failed) >> PAGE_SHIFT);
Signature · 5 lines
Kees Cook _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 6/7] i915: Convert to verify_page_range()
Quoted 13 lines · Kees Cook
drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c static int check_absent_pte(pte_t *pte, · line 1253
Quoted 8 lines · Peter Zijlstra
drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c +2
-2@@ -1249,14 +1249,14 @@ static int check_absent_pte(pte_t *pte,static int check_present (unsigned long addr,unsigned long len) { -return apply_to_page_range (current->mm, addr, len, - check_present_pte, (void *)addr); +return verify_page_range (current->mm, addr, len, + check_present_pte, (void *)addr);For example, switch to returning bad addr through
verify_page_range(), or have a by-reference value, etc:
Quoted 6 lines
unsigned long failed; failed = verify_page_range(current->mm< addr, len, check_present_pte); if (failed) {pr_err("missing PTE:%lx\n",
(addr - failed) >>PAGE_SHIFT);
OK, lemme try that.
Re: [PATCH 6/7] i915: Convert to verify_page_range()
In reply to Peter ZijlstraQuoted 35 lines · Peter Zijlstra
check_{present,absent}() only need R/O access, use
verify_page_range()instead to remove modular use ofapply_to_page_range().Signed-off-by Peter Zijlstra (Intel) <[email protected]>--- drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-)drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c +4
-4@@ -1225,9 +1225,9 @@ static int igt_mmap_gpu(void *arg)return 0 ; } -static int check_present_pte (pte_t *pte,unsigned long addr,void *data) +static int check_present_pte (pte_t pte,unsigned long addr,void *data) { -if (!pte_present (*pte) ||pte_none (*pte)) { +if (!pte_present (pte) ||pte_none (pte)) {pr_err ("missing PTE:%lx\n" , (addr - (unsigned long )data) >>PAGE_SHIFT );return -EINVAL ;@@ -1236,9 +1236,9 @@ static int check_present_pte(pte_t *pte,return 0 ; } -static int check_absent_pte (pte_t *pte,unsigned long addr,void *data) +static int check_absent_pte (pte_t pte,unsigned long addr,void *data) { -if (pte_present (*pte) && !pte_none (*pte)) { +if (pte_present (pte) && !pte_none (pte)) {pr_err ("present PTE:%lx; expected to be revoked\n" , (addr - (unsigned long )data) >>PAGE_SHIFT );return -EINVAL ;
drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c static int check_absent_pte(pte_t *pte, · line 1254
Quoted 8 lines
drivers/gpu/drm/i915/gem/selftests/i915_gem_mman.c +2
-2@@ -1249,14 +1249,14 @@ static int check_absent_pte(pte_t *pte,static int check_present (unsigned long addr,unsigned long len) { -return apply_to_page_range (current->mm, addr, len, - check_present_pte, (void *)addr); +return verify_page_range (current->mm, addr, len, + check_present_pte, (void *)addr); }
And this would be:
static int check_present(unsigned long addr, unsigned long len)
unsigned long fail;
fail = verify_page_range(current->mm, addr, len, check_present_pte);
if (fail) {
pr_err("missing PTE:%lx\n", addr);
return -EINVAL;
}
}(Oh, and I think I messed up the page shifting macro name in the earlier one...)
Signature · 1 line
Kees Cook
PATCH 7/7 mm: Unexport apply_to_page_range()
In reply to Peter ZijlstraNow that all module users of apply_to_page_range() have been removed, unexport this function.
This is an unsafe function in that it gives direct access to the page-tables.
--- mm/memory.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)
mm/memory.c +4 -1
@@ -2544,13 +2544,14 @@ static int __apply_to_page_range(struct/* * Scan a region of virtual memory, filling in page tables as necessary * and calling a provided function on each leaf page table. +* +* DO NOT EXPORT; this hands out our page-tables on a platter. */ int apply_to_page_range (struct mm_struct *mm,unsigned long addr,unsigned long size,pte_fn_t fn,void *data) {return __apply_to_page_range (mm, addr, size, fn, data, true); } -EXPORT_SYMBOL_GPL (apply_to_page_range);/* * Scan a region of virtual memory, calling a provided function on
@@ -2558,6 +2559,8 @@ EXPORT_SYMBOL_GPL(apply_to_page_range);* * Unlike apply_to_page_range, this does _not_ fill in page tables * where they are absent. +* +* DO NOT EXPORT; this hands out our page-tables on a platter. */ int apply_to_existing_page_range (struct mm_struct *mm,unsigned long addr,unsigned long size,pte_fn_t fn,void *data)
Looks good,
Patchwork
CI[Intel-gfx] ✗ Fi.CI.BUILD: failure for mm: Unexport apply_to_page_range()
In reply to Peter Zijlstra== Series Details ==
== Series Details ==
Series: mm: Unexport apply_to_page_range()
URL : https://patchwork.freedesktop.org/series/88943/
State : failure
== Summary ==
CALL scripts/checksyscalls.sh
CALL scripts/atomic/check-atomics.sh
DESCEND objtool
CHK include/generated/compile.h
Kernel: arch/x86/boot/bzImage is ready (#1)
MODPOST Module.symvers
ERROR: modpost: "apply_to_page_range" [drivers/gpu/drm/i915/i915.ko] undefined! scripts/Makefile.modpost:132: recipe for target 'Module.symvers' failed make[1]: *** [Module.symvers] Error 1
make[1]: *** Deleting file 'Module.symvers'
Makefile:1442: recipe for target 'modules' failed
make: *** [modules] Error 2
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [Intel-gfx] ✗ Fi.CI.BUILD: failure for mm: Unexport apply_to_page_range()
Quoted 8 lines · Patchwork
== Series Details ==
Series: mm: Unexport
apply_to_page_range()
URL : https://patchwork.freedesktop.org/series/88943/
State : failure== Summary ==
Quoted 7 lines
CALL
scripts/checksyscalls.sh
CALLscripts/atomic/check-atomics.sh
DESCEND objtool
CHKinclude/generated/compile.h
Kernel:arch/x86/boot/bzImageis ready (#1)
MODPOST Module.symvers
ERROR: modpost: "apply_to_page_range" [drivers/gpu/drm/i915/i915.ko] undefined!
This would be your robot not including the remap_pfn_range_notrack() patches from Christoph I suppose..
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx