clean up and streamline probe_kernel_* and friends v2
Subsystems (37)
Hi all,
this series start cleaning up the safe kernel and user memory probing helpers in mm/maccess.c, and then allows architectures to implement the kernel probing without overriding the address space limit and temporarily allowing access to user memory. It then switches x86 over to this new mechanism by reusing the unsafe_* uaccess logic.
This version also switches to the saner copy_{from,to}_kernel_nofault naming suggested by Linus.
I kept the x86 helprs as-is without calling unsage_{get,put}_user as that avoids a number of hard to trace casts, and it will still work with the asm-goto based version easily.
_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um
PATCH 1/18 maccess: unexport probe_kernel_write and probe_user_write
These two functions are not used by any modular code.
--- mm/maccess.c | 2 -- 1 file changed, 2 deletions(-)
mm/maccess.c +0 -2
@@ -121,7 +121,6 @@ long __probe_kernel_write(void *dst, const void *src, size_t size)return ret; } -EXPORT_SYMBOL_GPL (probe_kernel_write);/** * probe_user_write(): safely attempt to write to a user-space location
@@ -148,7 +147,6 @@ long __probe_user_write(void __user *dst, const void *src, size_t size)return ret; } -EXPORT_SYMBOL_GPL (probe_user_write);/** * strncpy_from_unsafe: - Copy a NUL terminated string from unsafe address.
Signature · 7 lines
2.26.2 _______________________________________________ linux-um mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-um
PATCH 2/18 maccess: remove various unused weak aliases
In reply to Christoph Hellwigmaccess tends to define lots of underscore prefixed symbols that then have other weak aliases. But except for two cases they are never actually used, so remove them.
--- include/linux/uaccess.h | 3 --- mm/maccess.c | 19 +++---------------- 2 files changed, 3 insertions(+), 19 deletions(-)
include/linux/uaccess.h +0 -3
@@ -324,7 +324,6 @@ extern long __probe_kernel_read(void *dst, const void *src, size_t size);* happens, handle that and return -EFAULT. */ extern long probe_user_read (void *dst,const void __user *src,size_t size); -extern long __probe_user_read (void *dst,const void __user *src,size_t size);/* * probe_kernel_write(): safely attempt to write to a location
@@ -336,7 +335,6 @@ extern long __probe_user_read(void *dst, const void __user *src, size_t size);* happens, handle that and return -EFAULT. */ extern long notrace probe_kernel_write (void *dst,const void *src,size_t size); -extern long notrace __probe_kernel_write (void *dst,const void *src,size_t size);/* * probe_user_write(): safely attempt to write to a location in user space
@@ -348,7 +346,6 @@ extern long notrace __probe_kernel_write(void *dst, const void *src, size_t size* happens, handle that and return -EFAULT. */ extern long notrace probe_user_write (void __user *dst,const void *src,size_t size); -extern long notrace __probe_user_write (void __user *dst,const void *src,size_t size);extern long strncpy_from_unsafe (char *dst,const void *unsafe_addr,long count); extern long strncpy_from_unsafe_strict(char *dst, const void *unsafe_addr,
mm/maccess.c +3 -16
@@ -79,11 +79,7 @@ EXPORT_SYMBOL_GPL(probe_kernel_read);* Safely read from user address @src to the buffer at @dst. If a kernel fault * happens, handle that and return -EFAULT. */ - -long __weak probe_user_read (void *dst,const void __user *src,size_t size) - __attribute__((alias ("__probe_user_read" ))); - -long __probe_user_read (void *dst,const void __user *src,size_t size) +long probe_user_read (void *dst,const void __user *src,size_t size) {long ret = -EFAULT ;mm_segment_t old_fs =get_fs ();
@@ -106,11 +102,7 @@ EXPORT_SYMBOL_GPL(probe_user_read);* Safely write to address @dst from the buffer at @src. If a kernel fault * happens, handle that and return -EFAULT. */ - -long __weak probe_kernel_write (void *dst,const void *src,size_t size) - __attribute__((alias ("__probe_kernel_write" ))); - -long __probe_kernel_write (void *dst,const void *src,size_t size) +long probe_kernel_write (void *dst,const void *src,size_t size) {long ret;mm_segment_t old_fs =get_fs ();
@@ -131,11 +123,7 @@ long __probe_kernel_write(void *dst, const void *src, size_t size)* Safely write to address @dst from the buffer at @src. If a kernel fault * happens, handle that and return -EFAULT. */ - -long __weak probe_user_write (void __user *dst,const void *src,size_t size) - __attribute__((alias ("__probe_user_write" ))); - -long __probe_user_write (void __user *dst,const void *src,size_t size) +long probe_user_write (void __user *dst,const void *src,size_t size) {long ret = -EFAULT ;mm_segment_t old_fs =get_fs ();
@@ -171,7 +159,6 @@ long __probe_user_write(void __user *dst, const void *src, size_t size)* probing memory on a user address range where strncpy_from_unsafe_user() is * supposed to be used instead. */ -long __weak strncpy_from_unsafe (char *dst,const void *unsafe_addr,long count) __attribute__((alias ("__strncpy_from_unsafe" )));
Signature · 7 lines
2.26.2 _______________________________________________ linux-um mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-um
PATCH 3/18 maccess: remove duplicate kerneldoc comments
In reply to Christoph HellwigMany of the maccess routines have a copy of the kerneldoc comment in the header. Remove it as it is not useful and will get out of sync sooner or later.
--- include/linux/uaccess.h | 38 -------------------------------------- 1 file changed, 38 deletions(-)
include/linux/uaccess.h +0 -38
@@ -301,50 +301,12 @@ copy_struct_from_user(void *dst, size_t ksize, const void __user *src,return 0 ; } -/* -* probe_kernel_read(): safely attempt to read from a location -* @dst: pointer to the buffer that shall take the data -* @src: address to read from -* @size: size of the data chunk -* -* Safely read from address @src to the buffer at @dst. If a kernel fault -* happens, handle that and return -EFAULT. -*/ extern long probe_kernel_read (void *dst,const void *src,size_t size);extern long probe_kernel_read_strict (void *dst,const void *src,size_t size);extern long __probe_kernel_read (void *dst,const void *src,size_t size); - -/* -* probe_user_read(): safely attempt to read from a location in user space -* @dst: pointer to the buffer that shall take the data -* @src: address to read from -* @size: size of the data chunk -* -* Safely read from address @src to the buffer at @dst. If a kernel fault -* happens, handle that and return -EFAULT. -*/ extern long probe_user_read (void *dst,const void __user *src,size_t size); -/* -* probe_kernel_write(): safely attempt to write to a location -* @dst: address to write to -* @src: pointer to the data that shall be written -* @size: size of the data chunk -* -* Safely write to address @dst from the buffer at @src. If a kernel fault -* happens, handle that and return -EFAULT. -*/ extern long notrace probe_kernel_write (void *dst,const void *src,size_t size); - -/* -* probe_user_write(): safely attempt to write to a location in user space -* @dst: address to write to -* @src: pointer to the data that shall be written -* @size: size of the data chunk -* -* Safely write to address @dst from the buffer at @src. If a kernel fault -* happens, handle that and return -EFAULT. -*/ extern long notrace probe_user_write (void __user *dst,const void *src,size_t size);extern long strncpy_from_unsafe (char *dst,const void *unsafe_addr,long count);
Signature · 7 lines
2.26.2 _______________________________________________ linux-um mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-um
PATCH 4/18 maccess: clarify kerneldoc comments
In reply to Christoph HellwigAdd proper kerneldoc comments for probe_kernel_read_strict and probe_kernel_read strncpy_from_unsafe_strict and explain the different versus the non-strict version.
--- mm/maccess.c | 61 ++++++++++++++++++++++++++++++++++++---------------- 1 file changed, 43 insertions(+), 18 deletions(-)
mm/maccess.c +43 -18
@@ -31,29 +31,35 @@ probe_write_common(void __user *dst, const void *src, size_t size) }/** -* probe_kernel_read(): safely attempt to read from a kernel-space location +* probe_kernel_read(): safely attempt to read from any location * @dst: pointer to the buffer that shall take the data * @src: address to read from * @size: size of the data chunk * -* Safely read from address @src to the buffer at @dst. If a kernel fault -* happens, handle that and return -EFAULT. +* Same as probe_kernel_read_strict() except that for architectures with +* not fully separated user and kernel address spaces this function also works +* for user address tanges. +* +* DO NOT USE THIS FUNCTION - it is broken on architectures with entirely +* separate kernel and user address spaces, and also a bad idea otherwise. +*/ +long __weak probe_kernel_read (void *dst,const void *src,size_t size) + __attribute__((alias ("__probe_kernel_read" ))); + +/** +* probe_kernel_read_strict(): safely attempt to read from kernel-space +* @dst: pointer to the buffer that shall take the data +* @src: address to read from +* @size: size of the data chunk +* +* Safely read from kernel address @src to the buffer at @dst. If a kernel +* fault happens, handle that and return -EFAULT. * * We ensure that the copy_from_user is executed in atomic context so that * do_page_fault() doesn't attempt to take mmap_sem. This makes * probe_kernel_read() suitable for use within regions where the caller * already holds mmap_sem, or other locks which nest inside mmap_sem. -* -* probe_kernel_read_strict() is the same as probe_kernel_read() except for -* the case where architectures have non-overlapping user and kernel address -* ranges: probe_kernel_read_strict() will additionally return -EFAULT for -* probing memory on a user address range where probe_user_read() is supposed -* to be used instead. */ - -long __weak probe_kernel_read (void *dst,const void *src,size_t size) - __attribute__((alias ("__probe_kernel_read" ))); -long __weak probe_kernel_read_strict (void *dst,const void *src,size_t size) __attribute__((alias ("__probe_kernel_read" )));
@@ -153,15 +159,34 @@ long probe_user_write(void __user *dst, const void *src, size_t size)* If @count is smaller than the length of the string, copies @count-1 bytes, * sets the last byte of @dst buffer to NUL and returns @count. * -* strncpy_from_unsafe_strict() is the same as strncpy_from_unsafe() except -* for the case where architectures have non-overlapping user and kernel address -* ranges: strncpy_from_unsafe_strict() will additionally return -EFAULT for -* probing memory on a user address range where strncpy_from_unsafe_user() is -* supposed to be used instead. +* Same as strncpy_from_unsafe_strict() except that for architectures with +* not fully separated user and kernel address spaces this function also works +* for user address tanges. +* +* DO NOT USE THIS FUNCTION - it is broken on architectures with entirely +* separate kernel and user address spaces, and also a bad idea otherwise. */ long __weak strncpy_from_unsafe (char *dst,const void *unsafe_addr,long count) __attribute__((alias ("__strncpy_from_unsafe" ))); +/** +* strncpy_from_unsafe_strict: - Copy a NUL terminated string from unsafe +* address. +* @dst: Destination address, in kernel space. This buffer must be at +* least @count bytes long. +* @unsafe_addr: Unsafe address. +* @count: Maximum number of bytes to copy, including the trailing NUL. +* +* Copies a NUL-terminated string from unsafe address to kernel buffer. +* +* On success, returns the length of the string INCLUDING the trailing NUL. +* +* If access fails, returns -EFAULT (some data may have been copied +* and the trailing NUL added). +* +* If @count is smaller than the length of the string, copies @count-1 bytes, +* sets the last byte of @dst buffer to NUL and returns @count. +*/ long __weak strncpy_from_unsafe_strict (char *dst,const void *unsafe_addr,long count) __attribute__((alias ("__strncpy_from_unsafe" )));
Signature · 7 lines
2.26.2 _______________________________________________ linux-um mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-um
PATCH 5/18 maccess: update the top of file comment
In reply to Christoph HellwigThis file now also contains several helpers for accessing user memory.
--- mm/maccess.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
mm/maccess.c +1 -1
@@ -1,6 +1,6 @@// SPDX-License-Identifier: GPL-2.0-only /* -* Access kernel memory without faulting. +* Access kernel or user memory without faulting. */ #include <linux/export.h> #include <linux/mm.h>
Signature · 7 lines
2.26.2 _______________________________________________ linux-um mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-um
PATCH 6/18 maccess: rename strncpy_from_unsafe_user to strncpy_from_user_nofault
In reply to Christoph HellwigThis matches the naming of strncpy_from_user, and also makes it more clear what the function is supposed to do.
--- include/linux/uaccess.h | 4 ++-- kernel/trace/bpf_trace.c | 2 +- kernel/trace/trace_kprobe.c | 2 +- mm/maccess.c | 4 ++-- 4 files changed, 6 insertions(+), 6 deletions(-)
include/linux/uaccess.h +2 -2
@@ -313,8 +313,8 @@ extern long strncpy_from_unsafe(char *dst, const void *unsafe_addr, long count);extern long strncpy_from_unsafe_strict (char *dst,const void *unsafe_addr,long count);extern long __strncpy_from_unsafe (char *dst,const void *unsafe_addr,long count); -extern long strncpy_from_unsafe_user (char *dst,const void __user *unsafe_addr, -long count); +long strncpy_from_user_nofault (char *dst,const void __user *unsafe_addr, +long count);extern long strnlen_unsafe_user (const void __user *unsafe_addr,long count);/**
kernel/trace/bpf_trace.c +1 -1
@@ -159,7 +159,7 @@ static const struct bpf_func_proto bpf_probe_read_user_proto = { BPF_CALL_3(bpf_probe_read_user_str, void *, dst, u32, size, const void __user *, unsafe_ptr) { -int ret =strncpy_from_unsafe_user (dst, unsafe_ptr, size); +int ret =strncpy_from_user_nofault (dst, unsafe_ptr, size);if (unlikely (ret <0 ))memset (dst,0 , size);
kernel/trace/trace_kprobe.c +1 -1
@@ -1268,7 +1268,7 @@ fetch_store_string_user(unsigned long addr, void *dest, void *base) __dest =get_loc_data (dest, base); - ret =strncpy_from_unsafe_user (__dest, uaddr, maxlen); + ret =strncpy_from_user_nofault (__dest, uaddr, maxlen);if (ret >=0 ) *(u32 *)dest =make_data_loc (ret, __dest - base);
mm/maccess.c +2 -2
@@ -215,7 +215,7 @@ long __strncpy_from_unsafe(char *dst, const void *unsafe_addr, long count) }/** -* strncpy_from_unsafe_user: - Copy a NUL terminated string from unsafe user +* strncpy_from_user_nofault: - Copy a NUL terminated string from unsafe user * address. * @dst: Destination address, in kernel space. This buffer must be at * least @count bytes long.
@@ -232,7 +232,7 @@ long __strncpy_from_unsafe(char *dst, const void *unsafe_addr, long count)* If @count is smaller than the length of the string, copies @count-1 bytes, * sets the last byte of @dst buffer to NUL and returns @count. */ -long strncpy_from_unsafe_user (char *dst,const void __user *unsafe_addr, +long strncpy_from_user_nofault (char *dst,const void __user *unsafe_addr,long count) {mm_segment_t old_fs =get_fs ();
Signature · 7 lines
2.26.2 _______________________________________________ linux-um mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-um
PATCH 7/18 maccess: rename strncpy_from_unsafe_strict to strncpy_from_kernel_nofault
In reply to Christoph HellwigThis matches the naming of strncpy_from_user_nofault, and also makes it more clear what the function is supposed to do.
--- arch/x86/mm/maccess.c | 2 +- include/linux/uaccess.h | 4 ++-- kernel/trace/bpf_trace.c | 2 +- mm/maccess.c | 6 +++--- 4 files changed, 7 insertions(+), 7 deletions(-)
arch/x86/mm/maccess.c +1 -1
@@ -34,7 +34,7 @@ long probe_kernel_read_strict(void *dst, const void *src, size_t size)return __probe_kernel_read (dst, src, size); } -long strncpy_from_unsafe_strict (char *dst,const void *unsafe_addr,long count) +long strncpy_from_kernel_nofault (char *dst,const void *unsafe_addr,long count) {if (unlikely (invalid_probe_range ((unsigned long )unsafe_addr)))return -EFAULT ;
include/linux/uaccess.h +2 -2
@@ -310,8 +310,8 @@ extern long notrace probe_kernel_write(void *dst, const void *src, size_t size);extern long notrace probe_user_write (void __user *dst,const void *src,size_t size);extern long strncpy_from_unsafe (char *dst,const void *unsafe_addr,long count); -extern long strncpy_from_unsafe_strict (char *dst,const void *unsafe_addr, -long count); +long strncpy_from_kernel_nofault (char *dst,const void *unsafe_addr, +long count);extern long __strncpy_from_unsafe (char *dst,const void *unsafe_addr,long count);long strncpy_from_user_nofault (char *dst,const void __user *unsafe_addr,long count);
kernel/trace/bpf_trace.c +1 -1
@@ -240,7 +240,7 @@ bpf_probe_read_kernel_str_common(void *dst, u32 size, const void *unsafe_ptr,* is returned that can be used for bpf_perf_event_output() et al. */ ret = compat ?strncpy_from_unsafe (dst, unsafe_ptr, size) : -strncpy_from_unsafe_strict (dst, unsafe_ptr, size); +strncpy_from_kernel_nofault (dst, unsafe_ptr, size);if (unlikely (ret <0 )) out:memset (dst,0 , size);
mm/maccess.c +3 -3
@@ -159,7 +159,7 @@ long probe_user_write(void __user *dst, const void *src, size_t size)* If @count is smaller than the length of the string, copies @count-1 bytes, * sets the last byte of @dst buffer to NUL and returns @count. * -* Same as strncpy_from_unsafe_strict() except that for architectures with +* Same as strncpy_from_kernel_nofault() except that for architectures with * not fully separated user and kernel address spaces this function also works * for user address tanges. *
@@ -170,7 +170,7 @@ long __weak strncpy_from_unsafe(char *dst, const void *unsafe_addr, long count) __attribute__((alias("__strncpy_from_unsafe")));/** -* strncpy_from_unsafe_strict: - Copy a NUL terminated string from unsafe +* strncpy_from_kernel_nofault: - Copy a NUL terminated string from unsafe * address. * @dst: Destination address, in kernel space. This buffer must be at * least @count bytes long.
@@ -187,7 +187,7 @@ long __weak strncpy_from_unsafe(char *dst, const void *unsafe_addr, long count)* If @count is smaller than the length of the string, copies @count-1 bytes, * sets the last byte of @dst buffer to NUL and returns @count. */ -long __weak strncpy_from_unsafe_strict (char *dst,const void *unsafe_addr, +long __weak strncpy_from_kernel_nofault (char *dst,const void *unsafe_addr,long count) __attribute__((alias ("__strncpy_from_unsafe" )));
Signature · 7 lines
2.26.2 _______________________________________________ linux-um mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-um
PATCH 8/18 maccess: rename strnlen_unsafe_user to strnlen_user_nofault
In reply to Christoph HellwigThis matches the naming of strnlen_user, and also makes it more clear what the function is supposed to do.
--- include/linux/uaccess.h | 2 +- kernel/trace/trace_kprobe.c | 2 +- mm/maccess.c | 4 ++-- 3 files changed, 4 insertions(+), 4 deletions(-)
include/linux/uaccess.h +1 -1
@@ -315,7 +315,7 @@ long strncpy_from_kernel_nofault(char *dst, const void *unsafe_addr,extern long __strncpy_from_unsafe (char *dst,const void *unsafe_addr,long count);long strncpy_from_user_nofault (char *dst,const void __user *unsafe_addr,long count); -extern long strnlen_unsafe_user (const void __user *unsafe_addr,long count); +long strnlen_user_nofault (const void __user *unsafe_addr,long count);/** * probe_kernel_address(): safely attempt to read from a location
kernel/trace/trace_kprobe.c +1 -1
@@ -1221,7 +1221,7 @@ fetch_store_strlen_user(unsigned long addr) {const void __user *uaddr = (__force const void __user *)addr; -return strnlen_unsafe_user (uaddr,MAX_STRING_SIZE ); +return strnlen_user_nofault (uaddr,MAX_STRING_SIZE ); }/*
mm/maccess.c +2 -2
@@ -258,7 +258,7 @@ long strncpy_from_user_nofault(char *dst, const void __user *unsafe_addr, }/** -* strnlen_unsafe_user: - Get the size of a user string INCLUDING final NUL. +* strnlen_user_nofault: - Get the size of a user string INCLUDING final NUL. * @unsafe_addr: The string to measure. * @count: Maximum count (including NUL) *
@@ -273,7 +273,7 @@ long strncpy_from_user_nofault(char *dst, const void __user *unsafe_addr,* Unlike strnlen_user, this can be used from IRQ handler etc. because * it disables pagefaults. */ -long strnlen_unsafe_user (const void __user *unsafe_addr,long count) +long strnlen_user_nofault (const void __user *unsafe_addr,long count) {mm_segment_t old_fs =get_fs ();int ret;
Signature · 7 lines
2.26.2 _______________________________________________ linux-um mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-um
PATCH 9/18 maccess: remove probe_read_common and probe_write_common
In reply to Christoph HellwigEach of the helpers has just two callers, which also different in dealing with kernel or userspace pointers. Just open code the logic in the callers.
--- mm/maccess.c | 63 ++++++++++++++++++++++++---------------------------- 1 file changed, 29 insertions(+), 34 deletions(-)
mm/maccess.c +29 -34
@@ -6,30 +6,6 @@#include <linux/mm.h> #include <linux/uaccess.h> -static __always_inline long -probe_read_common (void *dst,const void __user *src,size_t size) -{ -long ret; - -pagefault_disable (); - ret =__copy_from_user_inatomic (dst, src, size); -pagefault_enable (); - -return ret ? -EFAULT :0 ; -} - -static __always_inline long -probe_write_common (void __user *dst,const void *src,size_t size) -{ -long ret; - -pagefault_disable (); - ret =__copy_to_user_inatomic (dst, src, size); -pagefault_enable (); - -return ret ? -EFAULT :0 ; -} -/** * probe_kernel_read(): safely attempt to read from any location * @dst: pointer to the buffer that shall take the data
@@ -69,10 +45,15 @@ long __probe_kernel_read(void *dst, const void *src, size_t size)mm_segment_t old_fs =get_fs ();set_fs (KERNEL_DS ); - ret =probe_read_common (dst, (__force const void __user *)src, size); +pagefault_disable (); + ret =__copy_from_user_inatomic (dst, (__force const void __user *)src, + size); +pagefault_enable ();set_fs (old_fs); -return ret; +if (ret) +return -EFAULT ; +return 0 ; }EXPORT_SYMBOL_GPL (probe_kernel_read);
@@ -91,11 +72,16 @@ long probe_user_read(void *dst, const void __user *src, size_t size)mm_segment_t old_fs =get_fs ();set_fs (USER_DS ); -if (access_ok (src, size)) - ret =probe_read_common (dst, src, size); +if (access_ok (src, size)) { +pagefault_disable (); + ret =__copy_from_user_inatomic (dst, src, size); +pagefault_enable (); + }set_fs (old_fs); -return ret; +if (ret) +return -EFAULT ; +return 0 ; }EXPORT_SYMBOL_GPL (probe_user_read);
@@ -114,10 +100,14 @@ long probe_kernel_write(void *dst, const void *src, size_t size)mm_segment_t old_fs =get_fs ();set_fs (KERNEL_DS ); - ret =probe_write_common ((__force void __user *)dst, src, size); +pagefault_disable (); + ret =__copy_to_user_inatomic ((__force void __user *)dst, src, size); +pagefault_enable ();set_fs (old_fs); -return ret; +if (ret) +return -EFAULT ; +return 0 ; }/**
@@ -135,11 +125,16 @@ long probe_user_write(void __user *dst, const void *src, size_t size)mm_segment_t old_fs =get_fs ();set_fs (USER_DS ); -if (access_ok (dst, size)) - ret =probe_write_common (dst, src, size); +if (access_ok (dst, size)) { +pagefault_disable (); + ret =__copy_to_user_inatomic (dst, src, size); +pagefault_enable (); + }set_fs (old_fs); -return ret; +if (ret) +return -EFAULT ; +return 0 ; }/**
Signature · 7 lines
2.26.2 _______________________________________________ linux-um mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-um
PATCH 10/18 maccess: unify the probe kernel arch hooks
In reply to Christoph HellwigCurrently architectures have to override every routine that probes kernel memory, which includes a pure read and strcpy, both in strict and not strict variants. Just provide a single arch hooks instead to make sure all architectures cover all the cases.
--- arch/parisc/lib/memcpy.c | 13 ++++------- arch/um/kernel/maccess.c | 11 ++++----- arch/x86/mm/maccess.c | 18 ++++----------- include/linux/uaccess.h | 6 +++-- mm/maccess.c | 49 ++++++++++++++++++++++++++++++---------- 5 files changed, 56 insertions(+), 41 deletions(-)
arch/parisc/lib/memcpy.c +5 -8
@@ -57,14 +57,11 @@ void * memcpy(void * dst,const void *src, size_t count)EXPORT_SYMBOL (raw_copy_in_user);EXPORT_SYMBOL (memcpy); -long probe_kernel_read (void *dst,const void *src,size_t size) +bool probe_kernel_read_allowed (void *dst,const void *unsafe_src,size_t size, +bool strict) { -unsigned long addr = (unsigned long )src; - -if (addr <PAGE_SIZE ) -return -EFAULT ; - +if ((unsigned long )unsafe_src <PAGE_SIZE ) +return false;/* check for I/O space F_EXTEND(0xfff00000) access as well? */ - -return __probe_kernel_read (dst, src, size); +return true; }
arch/um/kernel/maccess.c +5 -6
@@ -7,15 +7,14 @@#include <linux/kernel.h> #include <os.h> -long probe_kernel_read (void *dst,const void *src,size_t size) +bool probe_kernel_read_allowed (void *dst,const void *src,size_t size, +bool strict) {void *psrc = (void *)rounddown ((unsigned long )src,PAGE_SIZE );if ((unsigned long )src <PAGE_SIZE || size <=0 ) -return -EFAULT ; - +return false;if (os_mincore (psrc, size + src - psrc) <=0 ) -return -EFAULT ; - -return __probe_kernel_read (dst, src, size); +return false; +return true; }
arch/x86/mm/maccess.c +5 -13
@@ -26,18 +26,10 @@ static __always_inline bool invalid_probe_range(u64 vaddr) }#endif -long probe_kernel_read_strict (void *dst,const void *src,size_t size) +bool probe_kernel_read_allowed (void *dst,const void *unsafe_src,size_t size, +bool strict) { -if (unlikely (invalid_probe_range ((unsigned long )src))) -return -EFAULT ; - -return __probe_kernel_read (dst, src, size); -} - -long strncpy_from_kernel_nofault (char *dst,const void *unsafe_addr,long count) -{ -if (unlikely (invalid_probe_range ((unsigned long )unsafe_addr))) -return -EFAULT ; - -return __strncpy_from_unsafe (dst, unsafe_addr, count); +if (!strict) +return true; +return !invalid_probe_range ((unsigned long )unsafe_src); }
include/linux/uaccess.h +4 -2
@@ -301,9 +301,11 @@ copy_struct_from_user(void *dst, size_t ksize, const void __user *src,return 0 ; } +bool probe_kernel_read_allowed (void *dst,const void *unsafe_src, +size_t size,bool strict); +extern long probe_kernel_read (void *dst,const void *src,size_t size);extern long probe_kernel_read_strict (void *dst,const void *src,size_t size); -extern long __probe_kernel_read (void *dst,const void *src,size_t size);extern long probe_user_read (void *dst,const void __user *src,size_t size);extern long notrace probe_kernel_write (void *dst,const void *src,size_t size);
@@ -312,7 +314,7 @@ extern long notrace probe_user_write(void __user *dst, const void *src, size_t sextern long strncpy_from_unsafe (char *dst,const void *unsafe_addr,long count);long strncpy_from_kernel_nofault (char *dst,const void *unsafe_addr,long count); -extern long __strncpy_from_unsafe (char *dst,const void *unsafe_addr,long count); +long strncpy_from_user_nofault (char *dst,const void __user *unsafe_addr,long count);long strnlen_user_nofault (const void __user *unsafe_addr,long count);
mm/maccess.c +37 -12
97 lines not on this page · raw message
Signature · 7 lines
2.26.2 _______________________________________________ linux-um mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-um
Masami Hiramatsu
maintainerRe: [PATCH 10/18] maccess: unify the probe kernel arch hooks
Hi Christoph,
Quoted 10 lines · On Wed, 13 May 2020 18:00:30 +0200 Christoph Hellwig <[email protected]> wrote:
mm/maccess.c +4
-2@@ -36,14 +50,20 @@ long __weak probe_kernel_read(void *dst, const void *src, size_t size)* probe_kernel_read() suitable for use within regions where the caller * already holds mmap_sem, or other locks which nest inside mmap_sem. */ -long __weak probe_kernel_read_strict (void *dst,const void *src,size_t size) - __attribute__((alias ("__probe_kernel_read" ))); +long probe_kernel_read_strict (void *dst,const void *src,size_t size) +{ +return __probe_kernel_read (dst, src, size, true); +}
mm/maccess.c long __weak probe_kernel_read(void *dst, const void *src, size_t size) · line 65
Quoted 10 lines
mm/maccess.c +4
-1-long __probe_kernel_read (void *dst,const void *src,size_t size) +static long __probe_kernel_read (void *dst,const void *src,size_t size, +bool strict) {long ret;mm_segment_t old_fs =get_fs (); +if (!probe_kernel_read_allowed (dst, src, size, strict)) +return -EFAULT ;
Could you make this return -ERANGE instead of -EFAULT so that the caller can notice that the address might be user space?
Thank you,
Signature · 1 line
Masami Hiramatsu <[email protected]>
Re: [PATCH 10/18] maccess: unify the probe kernel arch hooks
Quoted 10 lines · Masami Hiramatsu
mm/maccess.c long __weak probe_kernel_read(void *dst, const void *src, size_t size) · line 65
Quoted 7 lines
mm/maccess.c +3
-0+ bool strict) {long ret;mm_segment_t old_fs =get_fs (); +if (!probe_kernel_read_allowed (dst, src, size, strict)) +return -EFAULT ;Could you make this return
-ERANGEinstead of-EFAULTso that the caller can notice that the address might be user space?
That is clearly a behavior change, so I don't want to mix it into this patch. But I can add it as another patch at the end.
_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um
PATCH 11/18 maccess: remove strncpy_from_unsafe
In reply to Christoph HellwigAll three callers really should try the explicit kernel and user copies instead. One has already deprecated the somewhat dangerous either kernel or user address concept, the other two still need to follow up eventually.
--- include/linux/uaccess.h | 1 - kernel/trace/bpf_trace.c | 39 +++++++++++++++++++++++++------------ kernel/trace/trace_kprobe.c | 5 ++++- mm/maccess.c | 39 +------------------------------------ 4 files changed, 32 insertions(+), 52 deletions(-)
include/linux/uaccess.h +0 -1
@@ -311,7 +311,6 @@ extern long probe_user_read(void *dst, const void __user *src, size_t size);extern long notrace probe_kernel_write (void *dst,const void *src,size_t size);extern long notrace probe_user_write (void __user *dst,const void *src,size_t size); -extern long strncpy_from_unsafe (char *dst,const void *unsafe_addr,long count);long strncpy_from_kernel_nofault (char *dst,const void *unsafe_addr,long count);
kernel/trace/bpf_trace.c +27 -12
@@ -226,12 +226,14 @@ static __always_inline int bpf_probe_read_kernel_str_common(void *dst, u32 size, const void *unsafe_ptr, const bool compat) { +const void __user *user_ptr = (__force const void __user *)unsafe_ptr;int ret =security_locked_down (LOCKDOWN_BPF_READ );if (unlikely (ret <0 )) - goto out; + goto fail; +/* -* The strncpy_from_unsafe_*() call will likely not fill the entire +* The strncpy_from_*_nofault() calls will likely not fill the entire * buffer, but that's okay in this circumstance as we're probing * arbitrary memory anyway similar to bpf_probe_read_*() and might * as well probe the stack. Thus, memory is explicitly cleared
@@ -239,11 +241,16 @@ bpf_probe_read_kernel_str_common(void *dst, u32 size, const void *unsafe_ptr,* code altogether don't copy garbage; otherwise length of string * is returned that can be used for bpf_perf_event_output() et al. */ - ret = compat ?strncpy_from_unsafe (dst, unsafe_ptr, size) : -strncpy_from_kernel_nofault (dst, unsafe_ptr, size); -if (unlikely (ret <0 )) -out: -memset (dst,0 , size); + ret =strncpy_from_kernel_nofault (dst, unsafe_ptr, size); +if (unlikely (ret <0 )) { +if (compat) + ret =strncpy_from_user_nofault (dst, user_ptr, size); +if (unlikely (ret <0 )) + goto fail; + } +return 0 ; +fail: +memset (dst,0 , size);return ret; }
@@ -321,6 +328,17 @@ static const struct bpf_func_proto *bpf_get_probe_write_proto(void)return &bpf_probe_write_user_proto; } +#define BPF_STRNCPY_LEN 64 + +static void bpf_strncpy (char *buf,long unsafe_addr) +{ + buf[0 ] =0 ; +if (strncpy_from_kernel_nofault (buf, (void *)unsafe_addr, +BPF_STRNCPY_LEN )) +strncpy_from_user_nofault (buf, (void __user *)unsafe_addr, +BPF_STRNCPY_LEN ); +} +/* * Only limited trace_printk() conversion specifiers allowed: * %d %i %u %x %ld %li %lu %lx %lld %lli %llu %llx %p %s
@@ -332,7 +350,7 @@ BPF_CALL_5(bpf_trace_printk, char *, fmt, u32, fmt_size, u64, arg1,int mod[3 ] = {};int fmt_cnt =0 ;u64 unsafe_addr; -char buf[64 ]; +char buf[BPF_STRNCPY_LEN ];int i;/*
@@ -387,10 +405,7 @@ BPF_CALL_5(bpf_trace_printk, char *, fmt, u32, fmt_size, u64, arg1, arg3 = (long ) buf;break ; } - buf[0 ] =0 ; -strncpy_from_unsafe (buf, - (void *) (long ) unsafe_addr, -sizeof (buf)); +bpf_strncpy (buf, unsafe_addr); }continue ; }
kernel/trace/trace_kprobe.c +4 -1
@@ -1244,7 +1244,10 @@ fetch_store_string(unsigned long addr, void *dest, void *base)* Try to get string again, since the string can be changed while * probing. */ - ret =strncpy_from_unsafe (__dest, (void *)addr, maxlen); + ret =strncpy_from_kernel_nofault (__dest, (void *)addr, maxlen); +if (ret <0 ) + ret =strncpy_from_user_nofault (__dest, (void __user *)addr, + maxlen);if (ret >=0 ) *(u32 *)dest =make_data_loc (ret, __dest - base);
mm/maccess.c +1 -38
@@ -8,8 +8,6 @@static long __probe_kernel_read (void *dst,const void *src,size_t size,bool strict); -static long __strncpy_from_unsafe (char *dst,const void *unsafe_addr, -long count,bool strict);bool __weak probe_kernel_read_allowed (void *dst,const void *unsafe_src,size_t size,bool strict)
@@ -156,35 +154,6 @@ long probe_user_write(void __user *dst, const void *src, size_t size)return 0 ; } -/** -* strncpy_from_unsafe: - Copy a NUL terminated string from unsafe address. -* @dst: Destination address, in kernel space. This buffer must be at -* least @count bytes long. -* @unsafe_addr: Unsafe address. -* @count: Maximum number of bytes to copy, including the trailing NUL. -* -* Copies a NUL-terminated string from unsafe address to kernel buffer. -* -* On success, returns the length of the string INCLUDING the trailing NUL. -* -* If access fails, returns -EFAULT (some data may have been copied -* and the trailing NUL added). -* -* If @count is smaller than the length of the string, copies @count-1 bytes, -* sets the last byte of @dst buffer to NUL and returns @count. -* -* Same as strncpy_from_kernel_nofault() except that for architectures with -* not fully separated user and kernel address spaces this function also works -* for user address tanges. -* -* DO NOT USE THIS FUNCTION - it is broken on architectures with entirely -* separate kernel and user address spaces, and also a bad idea otherwise. -*/ -long strncpy_from_unsafe (char *dst,const void *unsafe_addr,long count) -{ -return __strncpy_from_unsafe (dst, unsafe_addr, count, false); -} -/** * strncpy_from_kernel_nofault: - Copy a NUL terminated string from unsafe * address.
@@ -204,12 +173,6 @@ long strncpy_from_unsafe(char *dst, const void *unsafe_addr, long count)* sets the last byte of @dst buffer to NUL and returns @count. */ long strncpy_from_kernel_nofault (char *dst,const void *unsafe_addr,long count) -{ -return __strncpy_from_unsafe (dst, unsafe_addr, count, true); -} - -static long __strncpy_from_unsafe (char *dst,const void *unsafe_addr, -long count,bool strict) {mm_segment_t old_fs =get_fs ();const void *src = unsafe_addr;
@@ -217,7 +180,7 @@ static long __strncpy_from_unsafe(char *dst, const void *unsafe_addr,if (unlikely (count <=0 ))return 0 ; -if (!probe_kernel_read_allowed (dst, unsafe_addr, count, strict)) +if (!probe_kernel_read_allowed (dst, unsafe_addr, count, true))return -EFAULT ;set_fs (KERNEL_DS );
Signature · 7 lines
2.26.2 _______________________________________________ linux-um mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-um
Re: [PATCH 11/18] maccess: remove strncpy_from_unsafe
kernel/trace/bpf_trace.c static const struct bpf_func_proto *bpf_get_probe_write_proto(void) · line 340
Quoted 9 lines · Christoph Hellwig
kernel/trace/bpf_trace.c +8
-0+static void bpf_strncpy (char *buf,long unsafe_addr) +{ + buf[0 ] =0 ; +if (strncpy_from_kernel_nofault (buf, (void *)unsafe_addr, +BPF_STRNCPY_LEN )) +strncpy_from_user_nofault (buf, (void __user *)unsafe_addr, +BPF_STRNCPY_LEN ); +}
This seems buggy when I look at it.
It seems to think that strncpy_from_kernel_nofault() returns an error code.
Not so, unless I missed where you changed the rules.
It returns the length of the string for a successful copy. 0 is actually an error case (for count being <= 0).
So the test for success seems entirely wrong.
Also, I do wonder if we shouldn't gate this on TASK_SIZE, and do the user trial first. On architectures where this thing is valid in the first place (ie kernel and user addresses are separate), the test for address size would allow us to avoid a pointless fault due to an invalid kernel access to user space.
So I think this function should look something like
static void bpf_strncpy(char *buf, long unsafe_addr)
{
/* Try user address */
if (unsafe_addr < TASK_SIZE) {
void __user *ptr = (void __user *)unsafe_addr;
if (strncpy_from_user_nofault(buf, ptr, BPF_STRNCPY_LEN) >= 0)
return;
} /* .. fall back on trying kernel access */
buf[0] = 0;
strncpy_from_kernel_nofault(buf, (void *)unsafe_addr,BPF_STRNCPY_LEN);
}
or similar. No?
Linus
Re: [PATCH 11/18] maccess: remove strncpy_from_unsafe
Quoted 13 lines · Linus Torvalds
kernel/trace/bpf_trace.c static const struct bpf_func_proto *bpf_get_probe_write_proto(void) · line 340
Quoted 9 lines · Christoph Hellwig
kernel/trace/bpf_trace.c +8
-0+static void bpf_strncpy (char *buf,long unsafe_addr) +{ + buf[0 ] =0 ; +if (strncpy_from_kernel_nofault (buf, (void *)unsafe_addr, +BPF_STRNCPY_LEN )) +strncpy_from_user_nofault (buf, (void __user *)unsafe_addr, +BPF_STRNCPY_LEN ); +}This seems buggy when I look at it.
Quoted 3 lines
It seems to think that
strncpy_from_kernel_nofault()returns an error code.Not so, unless I missed where you changed the rules.
I didn't change the rules, so yes, this is wrong.
Quoted 17 lines
Also, I do wonder if we shouldn't gate this on
TASK_SIZE, and do the user trial first. On architectures where this thing is valid in the first place (ie kernel and user addresses are separate), the test for address size would allow us to avoid a pointless fault due to an invalid kernel access to user space.So I think this function should look something like
static void bpf_strncpy(char *buf, long unsafe_addr) { /* Try user address */ if (unsafe_addr < TASK_SIZE) { void __user *ptr = (void __user *)unsafe_addr; if (strncpy_from_user_nofault(buf, ptr, BPF_STRNCPY_LEN) >= 0) return; }
Quoted 7 lines
/* .. fall back on trying kernel access */ buf[0] = 0; strncpy_from_kernel_nofault(buf, (void *)unsafe_addr,
BPF_STRNCPY_LEN);
}or similar. No?
So on say s390 TASK_SIZE_USUALLy is (-PAGE_SIZE), which means we'd alway try the user copy first, which seems odd.
I'd really like to here from the bpf folks what the expected use case is here, and if the typical argument is kernel or user memory.
Daniel Borkmann
maintainerRe: [PATCH 11/18] maccess: remove strncpy_from_unsafe
Quoted 43 lines · Christoph Hellwig
Quoted 13 lines · Linus Torvalds
kernel/trace/bpf_trace.c static const struct bpf_func_proto *bpf_get_probe_write_proto(void) · line 340
Quoted 9 lines · Christoph Hellwig
kernel/trace/bpf_trace.c +8
-0+static void bpf_strncpy (char *buf,long unsafe_addr) +{ + buf[0 ] =0 ; +if (strncpy_from_kernel_nofault (buf, (void *)unsafe_addr, +BPF_STRNCPY_LEN )) +strncpy_from_user_nofault (buf, (void __user *)unsafe_addr, +BPF_STRNCPY_LEN ); +}This seems buggy when I look at it.
Quoted 3 lines
It seems to think that
strncpy_from_kernel_nofault()returns an error code.Not so, unless I missed where you changed the rules.
I didn't change the rules, so yes, this is wrong.
Quoted 23 lines
Also, I do wonder if we shouldn't gate this on
TASK_SIZE, and do the user trial first. On architectures where this thing is valid in the first place (ie kernel and user addresses are separate), the test for address size would allow us to avoid a pointless fault due to an invalid kernel access to user space.So I think this function should look something like
static void bpf_strncpy(char *buf, long unsafe_addr) { /* Try user address */ if (unsafe_addr < TASK_SIZE) { void __user *ptr = (void __user *)unsafe_addr; if (strncpy_from_user_nofault(buf, ptr, BPF_STRNCPY_LEN) >= 0) return; }/* .. fall back on trying kernel access */ buf[0] = 0; strncpy_from_kernel_nofault(buf, (void *)unsafe_addr, BPF_STRNCPY_LEN); }
Quoted 7 lines
Quoted 1 line
or similar. No?
So on say s390
TASK_SIZE_USUALLyis (-PAGE_SIZE), which means we'd alway try the user copy first, which seems odd.I'd really like to here from the bpf folks what the expected use case is here, and if the typical argument is kernel or user memory.
It's used for both. Given this is enabled on pretty much all program types, my assumption would be that usage is still more often on kernel memory than user one.
Re: [PATCH 11/18] maccess: remove strncpy_from_unsafe
Quoted 2 lines · Daniel Borkmann
It's used for both.
Daniel, BPF real;ly needs to make up its mind about that.
You *cannot* use ti for both.
Yes, it happens to work on x86 and some other architectures.
But on other architectures, the exact same pointer value can be a kernel pointer or a user pointer.
Quoted 2 lines
Given this is enabled on pretty much all program types, my
assumption would be that usage is still more often on kernel memory than user one.
You need to pick one.
If you know it is a user pointer, use strncpy_from_user() (possibly with disable_pagefault() aka strncpy_from_user_nofault()).
And if you know it is a kernel pointer, use strncpy_from_unsafe() (aka strncpy_from_kernel_nofault()).
You really can't pick the "randomly one or the other guess what I mean " option.
Linus
_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um
Daniel Borkmann
maintainerRe: [PATCH 11/18] maccess: remove strncpy_from_unsafe
Quoted 4 lines · Linus Torvalds
Quoted 2 lines · Daniel Borkmann
It's used for both.
Quoted 8 lines
Daniel, BPF real;ly needs to make up its mind about that.
You *cannot* use ti for both.
Yes, it happens to work on x86 and some other architectures.
But on other architectures, the exact same pointer value can be a kernel pointer or a user pointer.
Right, it has the same issue as with the old probe helper. I was merely stating that there are existing users (on x86) out there that use it this way, even though broken generally.
Quoted 5 lines
Quoted 2 lines
Given this is enabled on pretty much all program types, my
assumption would be that usage is still more often on kernel memory than user one.You need to pick one.
Quoted 7 lines
If you know it is a user pointer, use
strncpy_from_user()(possibly withdisable_pagefault()akastrncpy_from_user_nofault()).And if you know it is a kernel pointer, use
strncpy_from_unsafe()(akastrncpy_from_kernel_nofault()).You really can't pick the "randomly one or the other guess what I mean " option.
My preference would be to have %s, %sK, %sU for bpf_trace_printk() where the latter two result in an explicit strncpy_from_kernel_nofault() or strncpy_from_user_nofault() choice while the %s is converted as per your suggestion and it would still allow for a grace period to convert existing users to the new variants, similar with what we did on the bpf_probe_read_kernel() and bpf_probe_read_user() helpers to get this sorted out.
Thanks,
Daniel
Masami Hiramatsu
maintainerRe: [PATCH 11/18] maccess: remove strncpy_from_unsafe
In reply to Daniel BorkmannQuoted 46 lines · On Thu, 14 May 2020 00:36:28 +0200 Daniel Borkmann <[email protected]> wrote:
Quoted 45 lines · Christoph Hellwig
Quoted 13 lines · Linus Torvalds
kernel/trace/bpf_trace.c static const struct bpf_func_proto *bpf_get_probe_write_proto(void) · line 340
Quoted 9 lines · Christoph Hellwig
kernel/trace/bpf_trace.c +8
-0+static void bpf_strncpy (char *buf,long unsafe_addr) +{ + buf[0 ] =0 ; +if (strncpy_from_kernel_nofault (buf, (void *)unsafe_addr, +BPF_STRNCPY_LEN )) +strncpy_from_user_nofault (buf, (void __user *)unsafe_addr, +BPF_STRNCPY_LEN ); +}This seems buggy when I look at it.
Quoted 3 lines
It seems to think that
strncpy_from_kernel_nofault()returns an error code.Not so, unless I missed where you changed the rules.
I didn't change the rules, so yes, this is wrong.
Quoted 24 lines
Also, I do wonder if we shouldn't gate this on
TASK_SIZE, and do the user trial first. On architectures where this thing is valid in the first place (ie kernel and user addresses are separate), the test for address size would allow us to avoid a pointless fault due to an invalid kernel access to user space.So I think this function should look something like
static void bpf_strncpy(char *buf, long unsafe_addr) { /* Try user address */ if (unsafe_addr < TASK_SIZE) { void __user *ptr = (void __user *)unsafe_addr; if (strncpy_from_user_nofault(buf, ptr, BPF_STRNCPY_LEN) >= 0) return; }/* .. fall back on trying kernel access */ buf[0] = 0; strncpy_from_kernel_nofault(buf, (void *)unsafe_addr, BPF_STRNCPY_LEN); }or similar. No?
Quoted 8 lines
Quoted 5 lines
So on say s390
TASK_SIZE_USUALLyis (-PAGE_SIZE), which means we'd alway try the user copy first, which seems odd.I'd really like to here from the bpf folks what the expected use case is here, and if the typical argument is kernel or user memory.
It's used for both. Given this is enabled on pretty much all program types, my assumption would be that usage is still more often on kernel memory than user one.
For trace_kprobe.c current order (kernel -> user fallback) is preferred because it has another function dedicated for user memory.
Thank you,
Signature · 1 line
Masami Hiramatsu <[email protected]>
Re: [PATCH 11/18] maccess: remove strncpy_from_unsafe
Quoted 4 lines · Masami Hiramatsu
For
trace_kprobe.ccurrent order (kernel -> user fallback) is preferred because it has another function dedicated for user memory.
Well, then it should just use the "strict" kernel-only one for the non-user memory.
But yes, if there are legacy interfaces, then we might want to say "these continue to work for the legacy case on platforms where we can tell which kind of pointer it is from the bit pattern".
But we should likely at least disallow it entirely on platforms where we really can't - or pick one hardcoded choice. On sparc, you really _have_ to specify one or the other.
Linus
_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um
Masami Hiramatsu
maintainerRe: [PATCH 11/18] maccess: remove strncpy_from_unsafe
Quoted 6 lines · On Wed, 13 May 2020 16:59:40 -0700 Linus Torvalds…
Quoted 4 lines · Masami Hiramatsu
For
trace_kprobe.ccurrent order (kernel -> user fallback) is preferred because it has another function dedicated for user memory.
Quoted 6 lines
Well, then it should just use the "strict" kernel-only one for the non-user memory.
But yes, if there are legacy interfaces, then we might want to say "these continue to work for the legacy case on platforms where we can tell which kind of pointer it is from the bit pattern".
Yes, that was why I changed my mind and send reviewed-by last time.
https://lore.kernel.org/bpf/[email protected]/
Quoted 3 lines
But we should likely at least disallow it entirely on platforms where we really can't - or pick one hardcoded choice. On sparc, you really _have_ to specify one or the other.
OK. BTW, is there any way to detect the kernel/user space overlap on memory layout statically? If there, I can do it. (I don't like "if (CONFIG_X86)" thing....)
Or, maybe we need CONFIG_ARCH_OVERLAP_ADDRESS_SPACE?
Thank you,
Signature · 1 line
Masami Hiramatsu <[email protected]>
Re: [PATCH 11/18] maccess: remove strncpy_from_unsafe
Quoted 9 lines · Masami Hiramatsu
Quoted 4 lines
But we should likely at least disallow it entirely on platforms where we really can't - or pick one hardcoded choice. On sparc, you really _have_ to specify one or the other.
OK. BTW, is there any way to detect the kernel/user space overlap on memory layout statically? If there, I can do it. (I don't like "if (
CONFIG_X86)" thing....)
Or, maybe we needCONFIG_ARCH_OVERLAP_ADDRESS_SPACE?
I think it would be better to have a CONFIG variable that architectures can just 'select' to show that they are ok with separate kernel and user addresses.
Because I don't think we have any way to say that right now as-is. You can probably come up with hacky ways to approximate it, ie something like
if (TASK_SIZE_MAX > PAGE_OFFSET)
.... they overlap ..
which would almost work, but..
Linus
Masami Hiramatsu
maintainerRe: [PATCH 11/18] maccess: remove strncpy_from_unsafe
Quoted 15 lines · On Wed, 13 May 2020 19:43:24 -0700 Linus Torvalds…
Quoted 9 lines · Masami Hiramatsu
Quoted 4 lines
But we should likely at least disallow it entirely on platforms where we really can't - or pick one hardcoded choice. On sparc, you really _have_ to specify one or the other.
OK. BTW, is there any way to detect the kernel/user space overlap on memory layout statically? If there, I can do it. (I don't like "if (
CONFIG_X86)" thing....)
Or, maybe we needCONFIG_ARCH_OVERLAP_ADDRESS_SPACE?I think it would be better to have a CONFIG variable that architectures can just 'select' to show that they are ok with separate kernel and user addresses.
Quoted 8 lines
Because I don't think we have any way to say that right now as-is. You can probably come up with hacky ways to approximate it, ie something like
if (
TASK_SIZE_MAX>PAGE_OFFSET)
.... they overlap ..which would almost work, but..
It seems TASK_SIZE_MAX is defined only on x86 and s390, what about comparing STACK_TOP_MAX with PAGE_OFFSET ?
Anyway, I agree that the best way is introducing a CONFIG.
Thank you,
Signature · 1 line
Masami Hiramatsu <[email protected]>
Daniel Borkmann
maintainerRe: [PATCH 11/18] maccess: remove strncpy_from_unsafe
Quoted 21 lines · Masami Hiramatsu
Quoted 19 lines · On Wed, 13 May 2020 19:43:24 -0700 Linus Torvalds…
Quoted 9 lines · Masami Hiramatsu
Quoted 4 lines
But we should likely at least disallow it entirely on platforms where we really can't - or pick one hardcoded choice. On sparc, you really _have_ to specify one or the other.
OK. BTW, is there any way to detect the kernel/user space overlap on memory layout statically? If there, I can do it. (I don't like "if (
CONFIG_X86)" thing....)
Or, maybe we needCONFIG_ARCH_OVERLAP_ADDRESS_SPACE?I think it would be better to have a CONFIG variable that architectures can just 'select' to show that they are ok with separate kernel and user addresses.
Because I don't think we have any way to say that right now as-is. You can probably come up with hacky ways to approximate it, ie something like
Quoted 8 lines
Quoted 4 lines
if (
TASK_SIZE_MAX>PAGE_OFFSET)
.... they overlap ..which would almost work, but..
It seems
TASK_SIZE_MAXis defined only on x86 and s390, what about comparingSTACK_TOP_MAXwithPAGE_OFFSET?
Anyway, I agree that the best way is introducing a CONFIG.
Agree, CONFIG knob that archs can select feels cleanest. Fwiw, I've cooked up fixes for bpf side locally here and finishing up testing, will push out later today.
Thanks,
Daniel
Re: [PATCH 11/18] maccess: remove strncpy_from_unsafe
In reply to Daniel BorkmannQuoted 8 lines · Daniel Borkmann
Quoted 5 lines
So on say s390
TASK_SIZE_USUALLyis (-PAGE_SIZE), which means we'd alway try the user copy first, which seems odd.I'd really like to here from the bpf folks what the expected use case is here, and if the typical argument is kernel or user memory.
It's used for both. Given this is enabled on pretty much all program types, my assumption would be that usage is still more often on kernel memory than user one.
Then it needs an argument telling it which one to use. Look at sparc64. Or s390. Or parisc. Et sodding cetera.
The underlying model is that the kernel lives in a separate address space. Yes, on x86 it's actually sharing the page tables with userland, but that's not universal. The same address can be both a valid userland one _and_ a valid kernel one. You need to tell which one do you want.
Daniel Borkmann
maintainerRe: [PATCH 11/18] maccess: remove strncpy_from_unsafe
Quoted 11 lines · Al Viro
Quoted 8 lines · Daniel Borkmann
Quoted 5 lines
So on say s390
TASK_SIZE_USUALLyis (-PAGE_SIZE), which means we'd alway try the user copy first, which seems odd.I'd really like to here from the bpf folks what the expected use case is here, and if the typical argument is kernel or user memory.
It's used for both. Given this is enabled on pretty much all program types, my assumption would be that usage is still more often on kernel memory than user one.
Quoted 7 lines
Then it needs an argument telling it which one to use. Look at sparc64. Or s390. Or parisc. Et sodding cetera.
The underlying model is that the kernel lives in a separate address space. Yes, on x86 it's actually sharing the page tables with userland, but that's not universal. The same address can be both a valid userland one _and_ a valid kernel one. You need to tell which one do you want.
Yes, see also 6ae08ae3dea2 ("bpf: Add probe_read_{user, kernel} and probe_read_{user, kernel}_str helpers"), and my other reply wrt bpf_trace_printk() on how to address this. All I'm trying to say is that both bpf_probe_read() and bpf_trace_printk() do exist in this form since early [e]bpf days for ~5yrs now and while broken on non-x86 there are a lot of users on x86 for this in the wild, so they need to have a chance to migrate over to the new facilities before they are fully removed.
RE: [PATCH 11/18] maccess: remove strncpy_from_unsafe
From: Daniel Borkmann
Quoted 21 lines
Sent: 14 May 2020 00:59
Quoted 11 lines · Al Viro
Quoted 8 lines · Daniel Borkmann
Quoted 5 lines
So on say s390
TASK_SIZE_USUALLyis (-PAGE_SIZE), which means we'd alway try the user copy first, which seems odd.I'd really like to here from the bpf folks what the expected use case is here, and if the typical argument is kernel or user memory.
It's used for both. Given this is enabled on pretty much all program types, my assumption would be that usage is still more often on kernel memory than user one.
Quoted 7 lines
Then it needs an argument telling it which one to use. Look at sparc64. Or s390. Or parisc. Et sodding cetera.
The underlying model is that the kernel lives in a separate address space. Yes, on x86 it's actually sharing the page tables with userland, but that's not universal. The same address can be both a valid userland one _and_ a valid kernel one. You need to tell which one do you want.
Quoted 6 lines
Yes, see also
6ae08ae3dea2("bpf: Add probe_read_{user, kernel} and probe_read_{user, kernel}_str helpers"), and my other reply wrtbpf_trace_printk()on how to address this. All I'm trying to say is that bothbpf_probe_read()andbpf_trace_printk()do exist in this form since early [e]bpf days for ~5yrs now and while broken on non-x86 there are a lot of users on x86 for this in the wild, so they need to have a chance to migrate over to the new facilities before they are fully removed.
If it's not a stupid question why is a BPF program allowed to get into a situation where it might have an invalid kernel address.
It all stinks of a hole that allows all of kernel memory to be read and copied to userspace.
Now you might want to something special so that BPF programs just abort on OOPS instead of possibly paniking the kernel. But that is different from a copy that expects to be passed garbage.
David
-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)
Daniel Borkmann
maintainerRe: [PATCH 11/18] maccess: remove strncpy_from_unsafe
On 5/14/20 12:01 PM, David Laight wrote:
[...]
Quoted 9 lines
If it's not a stupid question why is a BPF program allowed to get into a situation where it might have an invalid kernel address.
It all stinks of a hole that allows all of kernel memory to be read and copied to userspace.
Now you might want to something special so that BPF programs just abort on OOPS instead of possibly paniking the kernel. But that is different from a copy that expects to be passed garbage.
I suggest you read up on probe_kernel_read() and its uses in tracing in general, looks like you haven't done that.
_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um
PATCH 12/18 maccess: always use strict semantics for probe_kernel_read
In reply to Christoph HellwigExcept for historical confusion in the kprobes/uprobes and bpf tracers there is no good reason to ever allow user memory accesses from probe_kernel_read. Make the tracers fall back to a probe_user_read if the probe_kernel_read falls to keep the core API clean.
--- arch/parisc/lib/memcpy.c | 3 +-- arch/um/kernel/maccess.c | 3 +-- arch/x86/mm/maccess.c | 5 +---- include/linux/uaccess.h | 4 +--- kernel/trace/bpf_trace.c | 20 +++++++++++++------ kernel/trace/trace_kprobe.c | 11 ++++++++++- mm/maccess.c | 39 ++++++------------------------------- 7 files changed, 34 insertions(+), 51 deletions(-)
arch/parisc/lib/memcpy.c +1 -2
@@ -57,8 +57,7 @@ void * memcpy(void * dst,const void *src, size_t count)EXPORT_SYMBOL (raw_copy_in_user);EXPORT_SYMBOL (memcpy); -bool probe_kernel_read_allowed (void *dst,const void *unsafe_src,size_t size, -bool strict) +bool probe_kernel_read_allowed (void *dst,const void *unsafe_src,size_t size) {if ((unsigned long )unsafe_src <PAGE_SIZE )return false;
arch/um/kernel/maccess.c +1 -2
@@ -7,8 +7,7 @@#include <linux/kernel.h> #include <os.h> -bool probe_kernel_read_allowed (void *dst,const void *src,size_t size, -bool strict) +bool probe_kernel_read_allowed (void *dst,const void *src,size_t size) {void *psrc = (void *)rounddown ((unsigned long )src,PAGE_SIZE );
arch/x86/mm/maccess.c +1 -4
@@ -26,10 +26,7 @@ static __always_inline bool invalid_probe_range(u64 vaddr) }#endif -bool probe_kernel_read_allowed (void *dst,const void *unsafe_src,size_t size, -bool strict) +bool probe_kernel_read_allowed (void *dst,const void *unsafe_src,size_t size) { -if (!strict) -return true;return !invalid_probe_range ((unsigned long )unsafe_src); }
include/linux/uaccess.h +1 -3
@@ -301,11 +301,9 @@ copy_struct_from_user(void *dst, size_t ksize, const void __user *src,return 0 ; } -bool probe_kernel_read_allowed (void *dst,const void *unsafe_src, -size_t size,bool strict); +bool probe_kernel_read_allowed (void *dst,const void *unsafe_src,size_t size);extern long probe_kernel_read (void *dst,const void *src,size_t size); -extern long probe_kernel_read_strict (void *dst,const void *src,size_t size);extern long probe_user_read (void *dst,const void __user *src,size_t size);extern long notrace probe_kernel_write (void *dst,const void *src,size_t size);
kernel/trace/bpf_trace.c +14 -6
@@ -180,15 +180,23 @@ static __always_inline int bpf_probe_read_kernel_common(void *dst, u32 size, const void *unsafe_ptr, const bool compat) { +const void __user *user_ptr = (__force const void __user *)unsafe_ptr;int ret =security_locked_down (LOCKDOWN_BPF_READ );if (unlikely (ret <0 )) - goto out; - ret = compat ?probe_kernel_read (dst, unsafe_ptr, size) : -probe_kernel_read_strict (dst, unsafe_ptr, size); -if (unlikely (ret <0 )) -out: -memset (dst,0 , size); + goto fail; + + ret =probe_kernel_read (dst, unsafe_ptr, size); +if (unlikely (ret <0 )) { +if (compat) + ret =probe_user_read (dst, user_ptr, size); +if (unlikely (ret <0 )) + goto fail; + } + +return 0 ; +fail: +memset (dst,0 , size);return ret; }
kernel/trace/trace_kprobe.c +10 -1
@@ -1209,6 +1209,9 @@ fetch_store_strlen(unsigned long addr)do { ret =probe_kernel_read (&c, (u8 *)addr + len,1 ); +if (ret) + ret =probe_user_read (&c, + (__force u8 __user *)addr + len,1 ); len++; }while (c && ret ==0 && len <MAX_STRING_SIZE );
@@ -1281,7 +1284,13 @@ fetch_store_string_user(unsigned long addr, void *dest, void *base) static nokprobe_inline intprobe_mem_read (void *dest,void *src,size_t size) { -return probe_kernel_read (dest, src, size); +const void __user *user_ptr = (__force const void __user *)src; +int ret; + + ret =probe_kernel_read (dest, src, size); +if (ret) + ret =probe_user_read (dest, user_ptr, size); +return ret; }static nokprobe_inline int
mm/maccess.c +6 -33
@@ -6,36 +6,14 @@#include <linux/mm.h> #include <linux/uaccess.h> -static long __probe_kernel_read (void *dst,const void *src,size_t size, -bool strict); -bool __weak probe_kernel_read_allowed (void *dst,const void *unsafe_src, -size_t size,bool strict) +size_t size) {return true; }/** -* probe_kernel_read(): safely attempt to read from any location -* @dst: pointer to the buffer that shall take the data -* @src: address to read from -* @size: size of the data chunk -* -* Same as probe_kernel_read_strict() except that for architectures with -* not fully separated user and kernel address spaces this function also works -* for user address tanges. -* -* DO NOT USE THIS FUNCTION - it is broken on architectures with entirely -* separate kernel and user address spaces, and also a bad idea otherwise. -*/ -long probe_kernel_read (void *dst,const void *src,size_t size) -{ -return __probe_kernel_read (dst, src, size, false); -} -EXPORT_SYMBOL_GPL (probe_kernel_read); - -/** -* probe_kernel_read_strict(): safely attempt to read from kernel-space +* probe_kernel_read(): safely attempt to read from kernel-space * @dst: pointer to the buffer that shall take the data * @src: address to read from * @size: size of the data chunk
@@ -48,18 +26,12 @@ EXPORT_SYMBOL_GPL(probe_kernel_read);* probe_kernel_read() suitable for use within regions where the caller * already holds mmap_sem, or other locks which nest inside mmap_sem. */ -long probe_kernel_read_strict (void *dst,const void *src,size_t size) -{ -return __probe_kernel_read (dst, src, size, true); -} - -static long __probe_kernel_read (void *dst,const void *src,size_t size, -bool strict) +long probe_kernel_read (void *dst,const void *src,size_t size) {long ret;mm_segment_t old_fs =get_fs (); -if (!probe_kernel_read_allowed (dst, src, size, strict)) +if (!probe_kernel_read_allowed (dst, src, size))return -EFAULT ;set_fs (KERNEL_DS );
@@ -73,6 +45,7 @@ static long __probe_kernel_read(void *dst, const void *src, size_t size,return -EFAULT ;return 0 ; } +EXPORT_SYMBOL_GPL (probe_kernel_read);/** * probe_user_read(): safely attempt to read from a user-space location
@@ -180,7 +153,7 @@ long strncpy_from_kernel_nofault(char *dst, const void *unsafe_addr, long count)if (unlikely (count <=0 ))return 0 ; -if (!probe_kernel_read_allowed (dst, unsafe_addr, count, true)) +if (!probe_kernel_read_allowed (dst, unsafe_addr, count))return -EFAULT ;set_fs (KERNEL_DS );
Signature · 7 lines
2.26.2 _______________________________________________ linux-um mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-um
PATCH 13/18 maccess: move user access routines together
In reply to Christoph HellwigMove kernel access vs user access routines together to ease upcoming ifdefs.
--- mm/maccess.c | 110 +++++++++++++++++++++++++-------------------------- 1 file changed, 55 insertions(+), 55 deletions(-)
mm/maccess.c +55 -55
@@ -47,34 +47,6 @@ long probe_kernel_read(void *dst, const void *src, size_t size) }EXPORT_SYMBOL_GPL (probe_kernel_read); -/** -* probe_user_read(): safely attempt to read from a user-space location -* @dst: pointer to the buffer that shall take the data -* @src: address to read from. This must be a user address. -* @size: size of the data chunk -* -* Safely read from user address @src to the buffer at @dst. If a kernel fault -* happens, handle that and return -EFAULT. -*/ -long probe_user_read (void *dst,const void __user *src,size_t size) -{ -long ret = -EFAULT ; -mm_segment_t old_fs =get_fs (); - -set_fs (USER_DS ); -if (access_ok (src, size)) { -pagefault_disable (); - ret =__copy_from_user_inatomic (dst, src, size); -pagefault_enable (); - } -set_fs (old_fs); - -if (ret) -return -EFAULT ; -return 0 ; -} -EXPORT_SYMBOL_GPL (probe_user_read); -/** * probe_kernel_write(): safely attempt to write to a location * @dst: address to write to
@@ -100,33 +72,6 @@ long probe_kernel_write(void *dst, const void *src, size_t size)return 0 ; } -/** -* probe_user_write(): safely attempt to write to a user-space location -* @dst: address to write to -* @src: pointer to the data that shall be written -* @size: size of the data chunk -* -* Safely write to address @dst from the buffer at @src. If a kernel fault -* happens, handle that and return -EFAULT. -*/ -long probe_user_write (void __user *dst,const void *src,size_t size) -{ -long ret = -EFAULT ; -mm_segment_t old_fs =get_fs (); - -set_fs (USER_DS ); -if (access_ok (dst, size)) { -pagefault_disable (); - ret =__copy_to_user_inatomic (dst, src, size); -pagefault_enable (); - } -set_fs (old_fs); - -if (ret) -return -EFAULT ; -return 0 ; -} -/** * strncpy_from_kernel_nofault: - Copy a NUL terminated string from unsafe * address.
@@ -170,6 +115,61 @@ long strncpy_from_kernel_nofault(char *dst, const void *unsafe_addr, long count)return ret ? -EFAULT : src - unsafe_addr; } +/** +* probe_user_read(): safely attempt to read from a user-space location +* @dst: pointer to the buffer that shall take the data +* @src: address to read from. This must be a user address. +* @size: size of the data chunk +* +* Safely read from user address @src to the buffer at @dst. If a kernel fault +* happens, handle that and return -EFAULT. +*/ +long probe_user_read (void *dst,const void __user *src,size_t size) +{ +long ret = -EFAULT ; +mm_segment_t old_fs =get_fs (); + +set_fs (USER_DS ); +if (access_ok (src, size)) { +pagefault_disable (); + ret =__copy_from_user_inatomic (dst, src, size); +pagefault_enable (); + } +set_fs (old_fs); + +if (ret) +return -EFAULT ; +return 0 ; +} +EXPORT_SYMBOL_GPL (probe_user_read); + +/** +* probe_user_write(): safely attempt to write to a user-space location +* @dst: address to write to +* @src: pointer to the data that shall be written +* @size: size of the data chunk +* +* Safely write to address @dst from the buffer at @src. If a kernel fault +* happens, handle that and return -EFAULT. +*/ +long probe_user_write (void __user *dst,const void *src,size_t size) +{ +long ret = -EFAULT ; +mm_segment_t old_fs =get_fs (); + +set_fs (USER_DS ); +if (access_ok (dst, size)) { +pagefault_disable (); + ret =__copy_to_user_inatomic (dst, src, size); +pagefault_enable (); + } +set_fs (old_fs); + +if (ret) +return -EFAULT ; +return 0 ; +} +/** * strncpy_from_user_nofault: - Copy a NUL terminated string from unsafe user * address.
Signature · 7 lines
2.26.2 _______________________________________________ linux-um mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-um
PATCH 14/18 maccess: allow architectures to provide kernel probing directly
In reply to Christoph HellwigProvide alternative versions of probe_kernel_read, probe_kernel_write and strncpy_from_kernel_unsafe that don't need set_fs magic, but instead use arch hooks that are modelled after unsafe_{get,put}_user to access kernel memory in an exception safe way.
--- mm/maccess.c | 76 ++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 76 insertions(+)
mm/maccess.c +76 -0
@@ -12,6 +12,81 @@ bool __weak probe_kernel_read_allowed(void *dst, const void *unsafe_src,return true; } +#ifdef HAVE_ARCH_PROBE_KERNEL + +#define probe_kernel_read_loop (dst, src, len, type, err_label) \ + while (len >= sizeof(type)) { \ + arch_kernel_read(dst, src, type, err_label); \ + dst += sizeof(type); \ + src += sizeof(type); \ + len -= sizeof(type); \ + } + +long probe_kernel_read (void *dst,const void *src,size_t size) +{ +if (!probe_kernel_read_allowed (dst, src, size)) +return -EFAULT ; + +pagefault_disable (); +probe_kernel_read_loop (dst, src, size, u64, Efault); +probe_kernel_read_loop (dst, src, size, u32, Efault); +probe_kernel_read_loop (dst, src, size, u16, Efault); +probe_kernel_read_loop (dst, src, size, u8, Efault); +pagefault_enable (); +return 0 ; +Efault: +pagefault_enable (); +return -EFAULT ; +} +EXPORT_SYMBOL_GPL (probe_kernel_read); + +#define probe_kernel_write_loop (dst, src, len, type, err_label) \ + while (len >= sizeof(type)) { \ + arch_kernel_write(dst, src, type, err_label); \ + dst += sizeof(type); \ + src += sizeof(type); \ + len -= sizeof(type); \ + } + +long probe_kernel_write (void *dst,const void *src,size_t size) +{ +pagefault_disable (); +probe_kernel_write_loop (dst, src, size, u64, Efault); +probe_kernel_write_loop (dst, src, size, u32, Efault); +probe_kernel_write_loop (dst, src, size, u16, Efault); +probe_kernel_write_loop (dst, src, size, u8, Efault); +pagefault_enable (); +return 0 ; +Efault: +pagefault_enable (); +return -EFAULT ; +} + +long strncpy_from_kernel_unsafe (char *dst,const void *unsafe_addr,long count) +{ +const void *src = unsafe_addr; + +if (unlikely (count <=0 )) +return 0 ; +if (!probe_kernel_read_allowed (dst, unsafe_addr, count)) +return -EFAULT ; + +pagefault_disable (); +do { +arch_kernel_read (dst, src, u8, Efault); + dst++; + src++; + }while (dst[-1 ] && src - unsafe_addr < count); +pagefault_enable (); + + dst[-1 ] ='\0' ; +return src - unsafe_addr; +Efault: +pagefault_enable (); + dst[-1 ] ='\0' ; +return -EFAULT ; +} +#else /* HAVE_ARCH_PROBE_KERNEL */ /** * probe_kernel_read(): safely attempt to read from kernel-space * @dst: pointer to the buffer that shall take the data
@@ -114,6 +189,7 @@ long strncpy_from_kernel_nofault(char *dst, const void *unsafe_addr, long count) return ret ? -EFAULT : src - unsafe_addr; } +#endif /* HAVE_ARCH_PROBE_KERNEL */ /** * probe_user_read(): safely attempt to read from a user-space location
Signature · 7 lines
2.26.2 _______________________________________________ linux-um mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-um
Re: [PATCH 14/18] maccess: allow architectures to provide kernel probing directly
Quoted 2 lines · Christoph Hellwig
+
arch_kernel_read(dst, src, type, err_label); \
I'm wondering if
(a) we shouldn't expose this as an interface in general
(b) it wouldn't be named differently..
The reason for (a) is that several users of the "copy_from_kernel_nofault()" interfaces just seem to want a single access from kernel mode.
The reason for (b) is that if we do expose this as a normal interface, it shouldn't be called "arch_kernel_read", and it should have the same semantics as "get_user_unsafe()".
IOW, maybe we should simply do exactly that: have a "get_kernel_nofault()" thing that looks exactly like unsafe_get_user().
On x86, it would basically be identical to unsafe_get_user().
And on architectures that only have the copy function, you'd just have a fallback something like this:
#define get_kernel_nofault(dst, src, err_label) do { \
typeof (*src) __gkn_result; \
if (probe_kernel_read(&__gkn_result, src) < 0) \
goto err_label; \
(dst) = __gkn_result; \} while (0)
and now the people who want to read a single kernel word can just do
get_kernel_nofault(n, untrusted_pointer, error);
and they're done.
And some day - when we get reliably "asm goto" wiith outputs - that "get_kernel_fault()" will literally be a single instruction asm with the proper exception handler marker, the way "put_user_unsafe()" already works (and the way "put_kernel_nofault()" would already work if it does the above).
Linus
Re: [PATCH 14/18] maccess: allow architectures to provide kernel probing directly
Quoted 5 lines · Linus Torvalds
Quoted 1 line
+
arch_kernel_read(dst, src, type, err_label); \I'm wondering if
(a) we shouldn't expose this as an interface in general
We do export something like it, currently it is called probe_kernel_address, and the last patch renames it to get_kernel_nofault. However it is implemented as a wrapper around probe_kernel_address / copy_from_kernel_nofault and thus not quite as efficient and without the magic goto semantics.
Quoted 1 line
(b) it wouldn't be named differently..
It probably should with all the renaming..
Re: [PATCH 14/18] maccess: allow architectures to provide kernel probing directly
Quoted 6 lines · Christoph Hellwig
We do export something like it, currently it is called
probe_kernel_address, and the last patch renames it toget_kernel_nofault. However it is implemented as a wrapper aroundprobe_kernel_address/copy_from_kernel_nofaultand thus not quite as efficient and without the magic goto semantics.
Looking at the current users of "probe_kernel_read()", it looks like it's almost mostly things that just want a single byte or word.
It's not 100% that: we definitely do several things that want the "copy" semantics vs the "get" semantics: on the x86 side we have CALL_INSN_SIZE and MAX_INSN_SIZE, and the ldttss_desc.
But the bulk of them do seem to be a single value.
I don't know if performance really matters here, but to me the whole "most users seem to want to read a single value" is what makes me think that maybe that should be the primary model, rather than have the copy model be the primary one and then we implement the single value case (badly) with a copy.
It probably doesn't matter that much. I certainly wouldn't hold this series up over it - it can be a future thing.
Linus
Re: [PATCH 14/18] maccess: allow architectures to provide kernel probing directly
Quoted 9 lines · Linus Torvalds
Looking at the current users of "
probe_kernel_read()", it looks like it's almost mostly things that just want a single byte or word.It's not 100% that: we definitely do several things that want the "copy" semantics vs the "get" semantics: on the x86 side we have
CALL_INSN_SIZEandMAX_INSN_SIZE, and theldttss_desc.But the bulk of them do seem to be a single value.
Quoted 8 lines
I don't know if performance really matters here, but to me the whole "most users seem to want to read a single value" is what makes me think that maybe that should be the primary model, rather than have the copy model be the primary one and then we implement the single value case (badly) with a copy.
It probably doesn't matter that much. I certainly wouldn't hold this series up over it - it can be a future thing.
I can make the get_kernel_nofault implementation suck a little less :)
Note that the arch helper (we could call it unsafe_get_kernel_nofault) we still need to have a pagefault_disable / pagefault_enable pair around the calls. So maybe keep the get_kernel_nofault interface as-is (without the goto label), and prepare the arch helpers for being used similar to unsafe_get_user once all architectures are converted. And I can throw in a few patches to convert callers from the copy semantics to the get semantics.
Masami Hiramatsu
maintainerRe: [PATCH 14/18] maccess: allow architectures to provide kernel probing directly
In reply to Christoph HellwigHi Christoph,
Quoted 4 lines · On Wed, 13 May 2020 18:00:34 +0200 Christoph Hellwig <[email protected]> wrote:
Provide alternative versions of
probe_kernel_read,probe_kernel_writeandstrncpy_from_kernel_unsafethat don't needset_fsmagic, but instead use arch hooks that are modelled after unsafe_{get,put}_user to access kernel memory in an exception safe way.
This patch seems to introduce new implementation of probe_kernel_read/write() and strncpy_from_kernel_unsafe(), but also drops copy_from/to_kernel_nofault() and strncpy_from_kernel_nofault() if HAVE_ARCH_PROBE_KERNEL is defined. In the result, this cause a link error with BPF and kprobe events.
BTW, what is the difference of *_unsafe() and *_nofault()? (maybe we make those to *_nofault() finally?)
Thank you,
Quoted 103 lines
Signed-off-by Christoph Hellwig <[email protected]>--- mm/maccess.c | 76 ++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 76 insertions(+)mm/maccess.c +76
-0@@ -12,6 +12,81 @@ bool __weak probe_kernel_read_allowed(void *dst, const void *unsafe_src,return true; } +#ifdef HAVE_ARCH_PROBE_KERNEL + +#define probe_kernel_read_loop (dst, src, len, type, err_label) \ + while (len >= sizeof(type)) { \ + arch_kernel_read(dst, src, type, err_label); \ + dst += sizeof(type); \ + src += sizeof(type); \ + len -= sizeof(type); \ + } + +long probe_kernel_read (void *dst,const void *src,size_t size) +{ +if (!probe_kernel_read_allowed (dst, src, size)) +return -EFAULT ; + +pagefault_disable (); +probe_kernel_read_loop (dst, src, size, u64, Efault); +probe_kernel_read_loop (dst, src, size, u32, Efault); +probe_kernel_read_loop (dst, src, size, u16, Efault); +probe_kernel_read_loop (dst, src, size, u8, Efault); +pagefault_enable (); +return 0 ; +Efault: +pagefault_enable (); +return -EFAULT ; +} +EXPORT_SYMBOL_GPL (probe_kernel_read); + +#define probe_kernel_write_loop (dst, src, len, type, err_label) \ + while (len >= sizeof(type)) { \ + arch_kernel_write(dst, src, type, err_label); \ + dst += sizeof(type); \ + src += sizeof(type); \ + len -= sizeof(type); \ + } + +long probe_kernel_write (void *dst,const void *src,size_t size) +{ +pagefault_disable (); +probe_kernel_write_loop (dst, src, size, u64, Efault); +probe_kernel_write_loop (dst, src, size, u32, Efault); +probe_kernel_write_loop (dst, src, size, u16, Efault); +probe_kernel_write_loop (dst, src, size, u8, Efault); +pagefault_enable (); +return 0 ; +Efault: +pagefault_enable (); +return -EFAULT ; +} + +long strncpy_from_kernel_unsafe (char *dst,const void *unsafe_addr,long count) +{ +const void *src = unsafe_addr; + +if (unlikely (count <=0 )) +return 0 ; +if (!probe_kernel_read_allowed (dst, unsafe_addr, count)) +return -EFAULT ; + +pagefault_disable (); +do { +arch_kernel_read (dst, src, u8, Efault); + dst++; + src++; + }while (dst[-1 ] && src - unsafe_addr < count); +pagefault_enable (); + + dst[-1 ] ='\0' ; +return src - unsafe_addr; +Efault: +pagefault_enable (); + dst[-1 ] ='\0' ; +return -EFAULT ; +} +#else /* HAVE_ARCH_PROBE_KERNEL */ /** * probe_kernel_read(): safely attempt to read from kernel-space * @dst: pointer to the buffer that shall take the data @@ -114,6 +189,7 @@ long strncpy_from_kernel_nofault(char *dst, const void *unsafe_addr, long count) return ret ? -EFAULT : src - unsafe_addr; } +#endif /* HAVE_ARCH_PROBE_KERNEL */ /** * probe_user_read(): safely attempt to read from a user-space location Signature · 1 line
2.26.2
Signature · 1 line
Masami Hiramatsu <[email protected]>
Re: [PATCH 14/18] maccess: allow architectures to provide kernel probing directly
Quoted 9 lines · Masami Hiramatsu
Quoted 4 lines
Provide alternative versions of
probe_kernel_read,probe_kernel_writeandstrncpy_from_kernel_unsafethat don't needset_fsmagic, but instead use arch hooks that are modelled after unsafe_{get,put}_user to access kernel memory in an exception safe way.This patch seems to introduce new implementation of probe_kernel_read/write() and
strncpy_from_kernel_unsafe(), but also drops copy_from/to_kernel_nofault() andstrncpy_from_kernel_nofault()ifHAVE_ARCH_PROBE_KERNELis defined. In the result, this cause a link error with BPF and kprobe events.
That was just a bug as I didn't commit the changes to switch everything to _nofault and remove _unsafe entirely, sorry.
PATCH 15/18 x86: use non-set_fs based maccess routines
In reply to Christoph HellwigProvide arch_kernel_read and arch_kernel_write routines to implement the maccess routines without messing with set_fs and without stac/clac that opens up access to user space.
--- arch/x86/include/asm/uaccess.h | 16 ++++++++++++++++ 1 file changed, 16 insertions(+)
arch/x86/include/asm/uaccess.h +16 -0
@@ -523,5 +523,21 @@ do { \unsafe_copy_loop (__ucu_dst, __ucu_src, __ucu_len, u8, label); \ } while (0) +#define HAVE_ARCH_PROBE_KERNEL + +#define arch_kernel_read (dst, src, type, err_label) \ +do { \ + int __kr_err; \ + \ + __get_user_size(*((type *)dst), (__force type __user *)src, \ + sizeof(type), __kr_err); \ + if (unlikely(__kr_err)) \ + goto err_label; \ +} while (0) + +#define arch_kernel_write (dst, src, type, err_label) \ + __put_user_size(*((type *)(src)), (__force type __user *)(dst), \ + sizeof(type), err_label) +#endif /* _ASM_X86_UACCESS_H */
Signature · 7 lines
2.26.2 _______________________________________________ linux-um mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-um
PATCH 16/18 maccess: rename probe_kernel_{read,write} to copy_{from,to}_kernel_nofault
In reply to Christoph HellwigBetter describe what these functions do.
--- arch/arm/kernel/ftrace.c | 3 +- arch/arm/kernel/kgdb.c | 2 +- arch/arm64/kernel/insn.c | 4 +-- arch/csky/kernel/ftrace.c | 5 ++-- arch/ia64/kernel/ftrace.c | 6 ++-- arch/mips/kernel/kprobes.c | 6 ++-- arch/nds32/kernel/ftrace.c | 5 ++-- arch/parisc/kernel/ftrace.c | 2 +- arch/parisc/kernel/kgdb.c | 4 +-- arch/parisc/lib/memcpy.c | 3 +- arch/powerpc/kernel/module_64.c | 5 ++-- arch/powerpc/kernel/trace/ftrace.c | 24 +++++++-------- arch/powerpc/perf/core-book3s.c | 3 +- arch/riscv/kernel/ftrace.c | 3 +- arch/riscv/kernel/patch.c | 4 +-- arch/s390/kernel/ftrace.c | 4 +-- arch/sh/kernel/ftrace.c | 6 ++-- arch/um/kernel/maccess.c | 2 +- arch/x86/include/asm/ptrace.h | 4 +-- arch/x86/kernel/dumpstack.c | 2 +- arch/x86/kernel/ftrace.c | 8 ++--- arch/x86/kernel/kgdb.c | 6 ++-- arch/x86/kernel/kprobes/core.c | 5 ++-- arch/x86/kernel/kprobes/opt.c | 2 +- arch/x86/kernel/traps.c | 3 +- arch/x86/mm/fault.c | 2 +- arch/x86/mm/init_32.c | 2 +- arch/x86/mm/maccess.c | 3 +- arch/x86/xen/enlighten_pv.c | 2 +- drivers/char/mem.c | 2 +- drivers/dio/dio.c | 6 ++-- drivers/input/serio/hp_sdc.c | 2 +- drivers/misc/kgdbts.c | 6 ++-- drivers/video/fbdev/hpfb.c | 2 +- fs/proc/kcore.c | 3 +- include/linux/uaccess.h | 14 +++++---- kernel/debug/debug_core.c | 6 ++-- kernel/debug/gdbstub.c | 6 ++-- kernel/debug/kdb/kdb_main.c | 3 +- kernel/debug/kdb/kdb_support.c | 7 +++-- kernel/kthread.c | 2 +- kernel/trace/bpf_trace.c | 2 +- kernel/trace/trace_kprobe.c | 4 +-- kernel/workqueue.c | 10 +++---- mm/maccess.c | 48 +++++++++++++++--------------- mm/rodata_test.c | 2 +- mm/slub.c | 2 +- 47 files changed, 137 insertions(+), 120 deletions(-)
arch/arm/kernel/ftrace.c +2 -1
@@ -84,7 +84,8 @@ static int ftrace_modify_code(unsigned long pc, unsigned long old, old =__opcode_to_mem_arm (old);if (validate) { -if (probe_kernel_read (&replaced, (void *)pc,MCOUNT_INSN_SIZE )) +if (copy_from_kernel_nofault (&replaced, (void *)pc, +MCOUNT_INSN_SIZE ))return -EFAULT ;if (replaced != old)
arch/arm/kernel/kgdb.c +1 -1
@@ -236,7 +236,7 @@ int kgdb_arch_set_breakpoint(struct kgdb_bkpt *bpt)/* patch_text() only supports int-sized breakpoints */ BUILD_BUG_ON (sizeof (int ) !=BREAK_INSTR_SIZE ); - err =probe_kernel_read (bpt->saved_instr, (char *)bpt->bpt_addr, + err =copy_from_kernel_nofault (bpt->saved_instr, (char *)bpt->bpt_addr,BREAK_INSTR_SIZE );if (err)return err;
arch/arm64/kernel/insn.c +2 -2
@@ -123,7 +123,7 @@ int __kprobes aarch64_insn_read(void *addr, u32 *insnp)int ret;__le32 val; - ret =probe_kernel_read (&val, addr,AARCH64_INSN_SIZE ); + ret =copy_from_kernel_nofault (&val, addr,AARCH64_INSN_SIZE );if (!ret) *insnp =le32_to_cpu (val);
@@ -139,7 +139,7 @@ static int __kprobes __aarch64_insn_write(void *addr, __le32 insn)raw_spin_lock_irqsave (&patch_lock, flags); waddr =patch_map (addr,FIX_TEXT_POKE0 ); - ret =probe_kernel_write (waddr, &insn,AARCH64_INSN_SIZE ); + ret =copy_to_kernel_nofault (waddr, &insn,AARCH64_INSN_SIZE );patch_unmap (FIX_TEXT_POKE0 );raw_spin_unlock_irqrestore (&patch_lock, flags);
arch/csky/kernel/ftrace.c +3 -2
@@ -72,7 +72,8 @@ static int ftrace_check_current_nop(unsigned long hook)uint16_t olds[7 ];unsigned long hook_pos = hook -2 ; -if (probe_kernel_read ((void *)olds, (void *)hook_pos,sizeof (nops))) +if (copy_from_kernel_nofault ((void *)olds, (void *)hook_pos, +sizeof (nops)))return -EFAULT ;if (memcmp ((void *)nops, (void *)olds,sizeof (nops))) {
@@ -97,7 +98,7 @@ static int ftrace_modify_code(unsigned long hook, unsigned long target,make_jbsr (target, hook, call, nolr); - ret =probe_kernel_write ((void *)hook_pos, enable ? call : nops, + ret =copy_to_kernel_nofault ((void *)hook_pos, enable ? call : nops,sizeof (nops));if (ret)return -EPERM ;
arch/ia64/kernel/ftrace.c +3 -3
@@ -108,7 +108,7 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, goto skip_check;/* read the text we want to modify */ -if (probe_kernel_read (replaced, (void *)ip,MCOUNT_INSN_SIZE )) +if (copy_from_kernel_nofault (replaced, (void *)ip,MCOUNT_INSN_SIZE ))return -EFAULT ;/* Make sure it is what we expect it to be */
@@ -117,7 +117,7 @@ ftrace_modify_code(unsigned long ip, unsigned char *old_code, skip_check:/* replace the text with the new text */ -if (probe_kernel_write (((void *)ip), new_code,MCOUNT_INSN_SIZE )) +if (copy_to_kernel_nofault (((void *)ip), new_code,MCOUNT_INSN_SIZE ))return -EPERM ;flush_icache_range (ip, ip +MCOUNT_INSN_SIZE );
@@ -129,7 +129,7 @@ static int ftrace_make_nop_check(struct dyn_ftrace *rec, unsigned long addr)unsigned char __attribute__((aligned (8 ))) replaced[MCOUNT_INSN_SIZE ];unsigned long ip = rec->ip; -if (probe_kernel_read (replaced, (void *)ip,MCOUNT_INSN_SIZE )) +if (copy_from_kernel_nofault (replaced, (void *)ip,MCOUNT_INSN_SIZE ))return -EFAULT ;if (rec->flags &FTRACE_FL_CONVERTED ) {struct ftrace_call_insn *call_insn, *tmp_call;
arch/mips/kernel/kprobes.c +3 -3
@@ -86,9 +86,9 @@ int __kprobes arch_prepare_kprobe(struct kprobe *p) goto out; } -if ((probe_kernel_read (&prev_insn, p->addr -1 , -sizeof (mips_instruction)) ==0 ) && -insn_has_delayslot (prev_insn)) { +if (copy_from_kernel_nofault (&prev_insn, p->addr -1 , +sizeof (mips_instruction)) ==0 && +insn_has_delayslot (prev_insn)) {pr_notice ("Kprobes for branch delayslot are not supported\n" ); ret = -EINVAL ; goto out;
arch/nds32/kernel/ftrace.c +3 -2
@@ -131,13 +131,14 @@ static int __ftrace_modify_code(unsigned long pc, unsigned long *old_insn,unsigned long orig_insn[3 ];if (validate) { -if (probe_kernel_read (orig_insn, (void *)pc,MCOUNT_INSN_SIZE )) +if (copy_from_kernel_nofault (orig_insn, (void *)pc, +MCOUNT_INSN_SIZE ))return -EFAULT ;if (memcmp (orig_insn, old_insn,MCOUNT_INSN_SIZE ))return -EINVAL ; } -if (probe_kernel_write ((void *)pc, new_insn,MCOUNT_INSN_SIZE )) +if (copy_to_kernel_nofault ((void *)pc, new_insn,MCOUNT_INSN_SIZE ))return -EPERM ;return 0 ;
arch/parisc/kernel/ftrace.c +1 -1
@@ -172,7 +172,7 @@ int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr) ip = (void *)(rec->ip +4 - size); - ret =probe_kernel_read (insn, ip, size); + ret =copy_from_kernel_nofault (insn, ip, size);if (ret)return ret;
arch/parisc/kernel/kgdb.c +2 -2
@@ -154,8 +154,8 @@ void kgdb_arch_set_pc(struct pt_regs *regs, unsigned long ip)int kgdb_arch_set_breakpoint (struct kgdb_bkpt *bpt) { -int ret =probe_kernel_read (bpt->saved_instr, (char *)bpt->bpt_addr, -BREAK_INSTR_SIZE ); +int ret =copy_from_kernel_nofault (bpt->saved_instr, + (char *)bpt->bpt_addr,BREAK_INSTR_SIZE );if (ret)return ret;
arch/parisc/lib/memcpy.c +2 -1
@@ -57,7 +57,8 @@ void * memcpy(void * dst,const void *src, size_t count)EXPORT_SYMBOL (raw_copy_in_user);EXPORT_SYMBOL (memcpy); -bool probe_kernel_read_allowed (void *dst,const void *unsafe_src,size_t size) +bool copy_from_kernel_nofault_allowed (void *dst,const void *unsafe_src, +size_t size) {if ((unsigned long )unsafe_src <PAGE_SIZE )return false;
arch/powerpc/kernel/module_64.c +3 -2
@@ -159,7 +159,7 @@ int module_trampoline_target(struct module *mod, unsigned long addr, stub = (struct ppc64_stub_entry *)addr; -if (probe_kernel_read (&magic, &stub->magic,sizeof (magic))) { +if (copy_from_kernel_nofault (&magic, &stub->magic,sizeof (magic))) {pr_err ("%s: fault reading magic for stub %lx for %s\n" , __func__, addr, mod->name);return -EFAULT ; }
@@ -169,7 +169,8 @@ int module_trampoline_target(struct module *mod, unsigned long addr,return -EFAULT ; } -if (probe_kernel_read (&funcdata, &stub->funcdata,sizeof (funcdata))) { +if (copy_from_kernel_nofault (&funcdata, &stub->funcdata, +sizeof (funcdata))) {pr_err ("%s: fault reading funcdata for stub %lx for %s\n" , __func__, addr, mod->name);return -EFAULT ; }
arch/powerpc/kernel/trace/ftrace.c +12 -12
@@ -67,7 +67,7 @@ ftrace_modify_code(unsigned long ip, unsigned int old, unsigned int new)*/ /* read the text we want to modify */ -if (probe_kernel_read (&replaced, (void *)ip,MCOUNT_INSN_SIZE )) +if (copy_from_kernel_nofault (&replaced, (void *)ip,MCOUNT_INSN_SIZE ))return -EFAULT ;/* Make sure it is what we expect it to be */
@@ -128,7 +128,7 @@ __ftrace_make_nop(struct module *mod,unsigned int op, pop;/* read where this goes */ -if (probe_kernel_read (&op, (void *)ip,sizeof (int ))) { +if (copy_from_kernel_nofault (&op, (void *)ip,sizeof (int ))) {pr_err ("Fetching opcode failed.\n" );return -EFAULT ; }
@@ -162,7 +162,7 @@ __ftrace_make_nop(struct module *mod,/* When using -mkernel_profile there is no load to jump over */ pop =PPC_INST_NOP ; -if (probe_kernel_read (&op, (void *)(ip -4 ),4 )) { +if (copy_from_kernel_nofault (&op, (void *)(ip -4 ),4 )) {pr_err ("Fetching instruction at %lx failed.\n" , ip -4 );return -EFAULT ; }
@@ -193,7 +193,7 @@ __ftrace_make_nop(struct module *mod,* Check what is in the next instruction. We can see ld r2,40(r1), but * on first pass after boot we will see mflr r0. */ -if (probe_kernel_read (&op, (void *)(ip+4 ),MCOUNT_INSN_SIZE )) { +if (copy_from_kernel_nofault (&op, (void *)(ip+4 ),MCOUNT_INSN_SIZE )) {pr_err ("Fetching op failed.\n" );return -EFAULT ; }
@@ -222,7 +222,7 @@ __ftrace_make_nop(struct module *mod,unsigned long ip = rec->ip;unsigned long tramp; -if (probe_kernel_read (&op, (void *)ip,MCOUNT_INSN_SIZE )) +if (copy_from_kernel_nofault (&op, (void *)ip,MCOUNT_INSN_SIZE ))return -EFAULT ;/* Make sure that that this is still a 24bit jump */
@@ -245,7 +245,7 @@ __ftrace_make_nop(struct module *mod,pr_devel ("ip:%lx jumps to %lx" , ip, tramp);/* Find where the trampoline jumps to */ -if (probe_kernel_read (jmp, (void *)tramp,sizeof (jmp))) { +if (copy_from_kernel_nofault (jmp, (void *)tramp,sizeof (jmp))) {pr_err ("Failed to read %lx\n" , tramp);return -EFAULT ; }
@@ -341,7 +341,7 @@ static int setup_mcount_compiler_tramp(unsigned long tramp)return -1 ;/* New trampoline -- read where this goes */ -if (probe_kernel_read (&op, (void *)tramp,sizeof (int ))) { +if (copy_from_kernel_nofault (&op, (void *)tramp,sizeof (int ))) {pr_debug ("Fetching opcode failed.\n" );return -1 ; }
@@ -391,7 +391,7 @@ static int __ftrace_make_nop_kernel(struct dyn_ftrace *rec, unsigned long addr)unsigned int op;/* Read where this goes */ -if (probe_kernel_read (&op, (void *)ip,sizeof (int ))) { +if (copy_from_kernel_nofault (&op, (void *)ip,sizeof (int ))) {pr_err ("Fetching opcode failed.\n" );return -EFAULT ; }
@@ -516,7 +516,7 @@ __ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)struct module *mod = rec->arch.mod;/* read where this goes */ -if (probe_kernel_read (op, ip,sizeof (op))) +if (copy_from_kernel_nofault (op, ip,sizeof (op)))return -EFAULT ;if (!expected_nop_sequence (ip, op[0 ], op[1 ])) {
@@ -578,7 +578,7 @@ __ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)unsigned long ip = rec->ip;/* read where this goes */ -if (probe_kernel_read (&op, (void *)ip,MCOUNT_INSN_SIZE )) +if (copy_from_kernel_nofault (&op, (void *)ip,MCOUNT_INSN_SIZE ))return -EFAULT ;/* It should be pointing to a nop */
@@ -634,7 +634,7 @@ static int __ftrace_make_call_kernel(struct dyn_ftrace *rec, unsigned long addr) }/* Make sure we have a nop */ -if (probe_kernel_read (&op, ip,sizeof (op))) { +if (copy_from_kernel_nofault (&op, ip,sizeof (op))) {pr_err ("Unable to read ftrace location %p\n" , ip);return -EFAULT ; }
@@ -712,7 +712,7 @@ __ftrace_modify_call(struct dyn_ftrace *rec, unsigned long old_addr, }/* read where this goes */ -if (probe_kernel_read (&op, (void *)ip,sizeof (int ))) { +if (copy_from_kernel_nofault (&op, (void *)ip,sizeof (int ))) {pr_err ("Fetching opcode failed.\n" );return -EFAULT ; }
arch/powerpc/perf/core-book3s.c +2 -1
@@ -418,7 +418,8 @@ static __u64 power_pmu_bhrb_to(u64 addr)__u64 target;if (is_kernel_addr (addr)) { -if (probe_kernel_read (&instr, (void *)addr,sizeof (instr))) +if (copy_from_kernel_nofault (&instr, (void *)addr, +sizeof (instr)))return 0 ;return branch_target (&instr);
arch/riscv/kernel/ftrace.c +2 -1
@@ -25,7 +25,8 @@ static int ftrace_check_current_call(unsigned long hook_pos,* Read the text we want to modify; * return must be -EFAULT on read error */ -if (probe_kernel_read (replaced, (void *)hook_pos,MCOUNT_INSN_SIZE )) +if (copy_from_kernel_nofault (replaced, (void *)hook_pos, +MCOUNT_INSN_SIZE ))return -EFAULT ;/*
arch/riscv/kernel/patch.c +2 -2
@@ -57,7 +57,7 @@ static int __kprobes riscv_insn_write(void *addr, const void *insn, size_t len) waddr =patch_map (addr,FIX_TEXT_POKE0 ); - ret =probe_kernel_write (waddr, insn, len); + ret =copy_to_kernel_nofault (waddr, insn, len);patch_unmap (FIX_TEXT_POKE0 );
@@ -71,7 +71,7 @@ static int __kprobes riscv_insn_write(void *addr, const void *insn, size_t len)#else static int __kprobes riscv_insn_write (void *addr,const void *insn,size_t len) { -return probe_kernel_write (addr, insn, len); +return copy_to_kernel_nofault (addr, insn, len); }#endif /* CONFIG_MMU */
arch/s390/kernel/ftrace.c +2 -2
@@ -99,7 +99,7 @@ int ftrace_make_nop(struct module *mod, struct dyn_ftrace *rec, {struct ftrace_insn orig, new, old; -if (probe_kernel_read (&old, (void *) rec->ip,sizeof (old))) +if (copy_from_kernel_nofault (&old, (void *) rec->ip,sizeof (old)))return -EFAULT ;if (addr ==MCOUNT_ADDR ) {/* Initial code replacement */
@@ -121,7 +121,7 @@ int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr) {struct ftrace_insn orig, new, old; -if (probe_kernel_read (&old, (void *) rec->ip,sizeof (old))) +if (copy_from_kernel_nofault (&old, (void *) rec->ip,sizeof (old)))return -EFAULT ;/* Replace nop with an ftrace call. */ ftrace_generate_nop_insn (&orig);
arch/sh/kernel/ftrace.c +3 -3
@@ -119,7 +119,7 @@ static void ftrace_mod_code(void)* But if one were to fail, then they all should, and if one were * to succeed, then they all should. */ - mod_code_status =probe_kernel_write (mod_code_ip, mod_code_newcode, + mod_code_status =copy_to_kernel_nofault (mod_code_ip, mod_code_newcode,MCOUNT_INSN_SIZE );/* if we fail, then kill any new writers */
@@ -203,7 +203,7 @@ static int ftrace_modify_code(unsigned long ip, unsigned char *old_code,*/ /* read the text we want to modify */ -if (probe_kernel_read (replaced, (void *)ip,MCOUNT_INSN_SIZE )) +if (copy_from_kernel_nofault (replaced, (void *)ip,MCOUNT_INSN_SIZE ))return -EFAULT ;/* Make sure it is what we expect it to be */
@@ -268,7 +268,7 @@ static int ftrace_mod(unsigned long ip, unsigned long old_addr, {unsigned char code[MCOUNT_INSN_SIZE ]; -if (probe_kernel_read (code, (void *)ip,MCOUNT_INSN_SIZE )) +if (copy_from_kernel_nofault (code, (void *)ip,MCOUNT_INSN_SIZE ))return -EFAULT ;if (old_addr !=__raw_readl ((unsigned long *)code))
arch/um/kernel/maccess.c +1 -1
@@ -7,7 +7,7 @@#include <linux/kernel.h> #include <os.h> -bool probe_kernel_read_allowed (void *dst,const void *src,size_t size) +bool copy_from_kernel_nofault_allowed (void *dst,const void *src,size_t size) {void *psrc = (void *)rounddown ((unsigned long )src,PAGE_SIZE );
arch/x86/include/asm/ptrace.h +2 -2
@@ -278,7 +278,7 @@ static inline unsigned long *regs_get_kernel_stack_nth_addr(struct pt_regs *regs }/* To avoid include hell, we can't include uaccess.h */ -extern long probe_kernel_read (void *dst,const void *src,size_t size); +extern long copy_from_kernel_nofault (void *dst,const void *src,size_t size);/** * regs_get_kernel_stack_nth() - get Nth entry of the stack
@@ -298,7 +298,7 @@ static inline unsigned long regs_get_kernel_stack_nth(struct pt_regs *regs, addr =regs_get_kernel_stack_nth_addr (regs, n);if (addr) { - ret =probe_kernel_read (&val, addr,sizeof (val)); + ret =copy_from_kernel_nofault (&val, addr,sizeof (val));if (!ret)return val; }
arch/x86/kernel/dumpstack.c +1 -1
@@ -106,7 +106,7 @@ void show_opcodes(struct pt_regs *regs, const char *loglvl) bad_ip =user_mode (regs) &&__chk_range_not_ok (prologue,OPCODE_BUFSIZE ,TASK_SIZE_MAX ); -if (bad_ip ||probe_kernel_read (opcodes, (u8 *)prologue, +if (bad_ip ||copy_from_kernel_nofault (opcodes, (u8 *)prologue,OPCODE_BUFSIZE )) {printk ("%sCode: Bad RIP value.\n" , loglvl); }else {
arch/x86/kernel/ftrace.c +4 -4
@@ -86,7 +86,7 @@ static int ftrace_verify_code(unsigned long ip, const char *old_code)* sure what we read is what we expected it to be before modifying it. */ /* read the text we want to modify */ -if (probe_kernel_read (cur_code, (void *)ip,MCOUNT_INSN_SIZE )) { +if (copy_from_kernel_nofault (cur_code, (void *)ip,MCOUNT_INSN_SIZE )) {WARN_ON (1 );return -EFAULT ; }
@@ -354,7 +354,7 @@ create_trampoline(struct ftrace_ops *ops, unsigned int *tramp_size) npages =DIV_ROUND_UP (*tramp_size,PAGE_SIZE );/* Copy ftrace_caller onto the trampoline memory */ - ret =probe_kernel_read (trampoline, (void *)start_offset, size); + ret =copy_from_kernel_nofault (trampoline, (void *)start_offset, size);if (WARN_ON (ret <0 )) goto fail;
@@ -362,7 +362,7 @@ create_trampoline(struct ftrace_ops *ops, unsigned int *tramp_size)/* The trampoline ends with ret(q) */ retq = (unsigned long )ftrace_stub; - ret =probe_kernel_read (ip, (void *)retq,RET_SIZE ); + ret =copy_from_kernel_nofault (ip, (void *)retq,RET_SIZE );if (WARN_ON (ret <0 )) goto fail;
@@ -471,7 +471,7 @@ static void *addr_from_call(void *ptr)union text_poke_insn call;int ret; - ret =probe_kernel_read (&call, ptr,CALL_INSN_SIZE ); + ret =copy_from_kernel_nofault (&call, ptr,CALL_INSN_SIZE );if (WARN_ON_ONCE (ret <0 ))return NULL ;
arch/x86/kernel/kgdb.c +3 -3
@@ -732,11 +732,11 @@ int kgdb_arch_set_breakpoint(struct kgdb_bkpt *bpt)int err; bpt->type =BP_BREAKPOINT ; - err =probe_kernel_read (bpt->saved_instr, (char *)bpt->bpt_addr, + err =copy_from_kernel_nofault (bpt->saved_instr, (char *)bpt->bpt_addr,BREAK_INSTR_SIZE );if (err)return err; - err =probe_kernel_write ((char *)bpt->bpt_addr, + err =copy_to_kernel_nofault ((char *)bpt->bpt_addr, arch_kgdb_ops.gdb_bpt_instr,BREAK_INSTR_SIZE );if (!err)return err;
@@ -768,7 +768,7 @@ int kgdb_arch_remove_breakpoint(struct kgdb_bkpt *bpt)return 0 ; knl_write: -return probe_kernel_write ((char *)bpt->bpt_addr, +return copy_to_kernel_nofault ((char *)bpt->bpt_addr, (char *)bpt->saved_instr,BREAK_INSTR_SIZE ); }
arch/x86/kernel/kprobes/core.c +3 -2
@@ -243,7 +243,7 @@ __recover_probed_insn(kprobe_opcode_t *buf, unsigned long addr)* Fortunately, we know that the original code is the ideal 5-byte * long NOP. */ -if (probe_kernel_read (buf, (void *)addr, +if (copy_from_kernel_nofault (buf, (void *)addr,MAX_INSN_SIZE *sizeof (kprobe_opcode_t)))return 0UL ;
@@ -346,7 +346,8 @@ int __copy_instruction(u8 *dest, u8 *src, u8 *real, struct insn *insn)return 0 ;/* This can access kernel text if given address is not recovered */ -if (probe_kernel_read (dest, (void *)recovered_insn,MAX_INSN_SIZE )) +if (copy_from_kernel_nofault (dest, (void *)recovered_insn, +MAX_INSN_SIZE ))return 0 ;kernel_insn_init (insn, dest,MAX_INSN_SIZE );
arch/x86/kernel/kprobes/opt.c +1 -1
@@ -56,7 +56,7 @@ unsigned long __recover_optprobed_insn(kprobe_opcode_t *buf, unsigned long addr)* overwritten by jump destination address. In this case, original * bytes must be recovered from op->optinsn.copied_insn buffer. */ -if (probe_kernel_read (buf, (void *)addr, +if (copy_from_kernel_nofault (buf, (void *)addr,MAX_INSN_SIZE *sizeof (kprobe_opcode_t)))return 0UL ;
arch/x86/kernel/traps.c +2 -1
@@ -483,7 +483,8 @@ static enum kernel_gp_hint get_kernel_gp_address(struct pt_regs *regs,u8 insn_buf[MAX_INSN_SIZE ];struct insn insn; -if (probe_kernel_read (insn_buf, (void *)regs->ip,MAX_INSN_SIZE )) +if (copy_from_kernel_nofault (insn_buf, (void *)regs->ip, +MAX_INSN_SIZE ))return GP_NO_HINT ;kernel_insn_init (&insn, insn_buf,MAX_INSN_SIZE );
arch/x86/mm/fault.c +1 -1
@@ -590,7 +590,7 @@ static void show_ldttss(const struct desc_ptr *gdt, const char *name, u16 index)return ; } -if (probe_kernel_read (&desc, (void *)(gdt->address + offset), +if (copy_from_kernel_nofault (&desc, (void *)(gdt->address + offset),sizeof (struct ldttss_desc ))) {pr_alert ("%s: 0x%hx -- GDT entry is not readable\n" , name, index);
arch/x86/mm/init_32.c +1 -1
@@ -758,7 +758,7 @@ static void __init test_wp_bit(void)__set_fixmap (FIX_WP_TEST ,__pa_symbol (empty_zero_page),PAGE_KERNEL_RO ); -if (probe_kernel_write ((char *)fix_to_virt (FIX_WP_TEST ), &z,1 )) { +if (copy_to_kernel_nofault ((char *)fix_to_virt (FIX_WP_TEST ), &z,1 )) {clear_fixmap (FIX_WP_TEST );printk (KERN_CONT "Ok.\n" );return ;
arch/x86/mm/maccess.c +2 -1
@@ -26,7 +26,8 @@ static __always_inline bool invalid_probe_range(u64 vaddr) }#endif -bool probe_kernel_read_allowed (void *dst,const void *unsafe_src,size_t size) +bool copy_from_kernel_nofault_allowed (void *dst,const void *unsafe_src, +size_t size) {return !invalid_probe_range ((unsigned long )unsafe_src); }
arch/x86/xen/enlighten_pv.c +1 -1
@@ -387,7 +387,7 @@ static void set_aliased_prot(void *v, pgprot_t prot)preempt_disable (); -probe_kernel_read (&dummy, v,1 ); +copy_from_kernel_nofault (&dummy, v,1 );if (HYPERVISOR_update_va_mapping ((unsigned long )v, pte,0 ))BUG ();
drivers/char/mem.c +1 -1
@@ -167,7 +167,7 @@ static ssize_t read_mem(struct file *file, char __user *buf,if (!ptr) goto failed; - probe =probe_kernel_read (bounce, ptr, sz); + probe =copy_from_kernel_nofault (bounce, ptr, sz);unxlate_dev_mem_ptr (p, ptr);if (probe) goto failed;
drivers/dio/dio.c +4 -2
@@ -135,7 +135,8 @@ int __init dio_find(int deviceid)else va =ioremap (pa,PAGE_SIZE ); -if (probe_kernel_read (&i, (unsigned char *)va +DIO_IDOFF ,1 )) { +if (copy_from_kernel_nofault (&i, + (unsigned char *)va +DIO_IDOFF ,1 )) {if (scode >=DIOII_SCBASE )iounmap (va);continue ;/* no board present at that select code */
@@ -208,7 +209,8 @@ static int __init dio_init(void)else va =ioremap (pa,PAGE_SIZE ); -if (probe_kernel_read (&i, (unsigned char *)va +DIO_IDOFF ,1 )) { +if (copy_from_kernel_nofault (&i, + (unsigned char *)va +DIO_IDOFF ,1 )) {if (scode >=DIOII_SCBASE )iounmap (va);continue ;/* no board present at that select code */
drivers/input/serio/hp_sdc.c +1 -1
@@ -1021,7 +1021,7 @@ static int __init hp_sdc_register(void) hp_sdc.base_io = (unsigned long )0xf0428000 ; hp_sdc.data_io = (unsigned long ) hp_sdc.base_io +1 ; hp_sdc.status_io = (unsigned long ) hp_sdc.base_io +3 ; -if (!probe_kernel_read (&i, (unsigned char *)hp_sdc.data_io,1 )) +if (!copy_from_kernel_nofault (&i, (unsigned char *)hp_sdc.data_io,1 )) hp_sdc.dev = (void *)1 ; hp_sdc.dev_err =hp_sdc_init ();#endif
drivers/misc/kgdbts.c +3 -3
@@ -828,7 +828,7 @@ static void run_plant_and_detach_test(int is_early)char before[BREAK_INSTR_SIZE ];char after[BREAK_INSTR_SIZE ]; -probe_kernel_read (before, (char *)kgdbts_break_test, +copy_from_kernel_nofault (before, (char *)kgdbts_break_test,BREAK_INSTR_SIZE );init_simple_test (); ts.tst = plant_and_detach_test;
@@ -836,8 +836,8 @@ static void run_plant_and_detach_test(int is_early)/* Activate test with initial breakpoint */ if (!is_early)kgdb_breakpoint (); -probe_kernel_read (after, (char *)kgdbts_break_test, -BREAK_INSTR_SIZE ); +copy_from_kernel_nofault (after, (char *)kgdbts_break_test, +BREAK_INSTR_SIZE );if (memcmp (before, after,BREAK_INSTR_SIZE )) {printk (KERN_CRIT "kgdbts: ERROR kgdb corrupted memory\n" );panic ("kgdb memory corruption" );
drivers/video/fbdev/hpfb.c +1 -1
@@ -402,7 +402,7 @@ int __init hpfb_init(void)if (err)return err; - err =probe_kernel_read (&i, (unsigned char *)INTFBVADDR +DIO_IDOFF ,1 ); + err =copy_from_kernel_nofault (&i, (unsigned char *)INTFBVADDR +DIO_IDOFF ,1 );if (!err && (i ==DIO_ID_FBUFFER ) &&topcat_sid_ok (sid =DIO_SECID (INTFBVADDR ))) {if (!request_mem_region (INTFBPADDR ,DIO_DEVSIZE ,"Internal Topcat" ))
fs/proc/kcore.c +2 -1
@@ -512,7 +512,8 @@ read_kcore(struct file *file, char __user *buffer, size_t buflen, loff_t *fpos)* Using bounce buffer to bypass the * hardened user copy kernel text checks. */ -if (probe_kernel_read (buf, (void *) start, tsz)) { +if (copy_from_kernel_nofault (buf, (void *)start, + tsz)) {if (clear_user (buffer, tsz)) { ret = -EFAULT ; goto out;
include/linux/uaccess.h +8 -6
@@ -301,13 +301,15 @@ copy_struct_from_user(void *dst, size_t ksize, const void __user *src,return 0 ; } -bool probe_kernel_read_allowed (void *dst,const void *unsafe_src,size_t size); +bool copy_from_kernel_nofault_allowed (void *dst,const void *unsafe_src, +size_t size); -extern long probe_kernel_read (void *dst,const void *src,size_t size); -extern long probe_user_read (void *dst,const void __user *src,size_t size); +long copy_from_kernel_nofault (void *dst,const void *src,size_t size); +long notrace copy_to_kernel_nofault (void *dst,const void *src,size_t size); -extern long notrace probe_kernel_write (void *dst,const void *src,size_t size); -extern long notrace probe_user_write (void __user *dst,const void *src,size_t size); +extern long probe_user_read (void *dst,const void __user *src,size_t size); +extern long notrace probe_user_write (void __user *dst,const void *src, +size_t size);long strncpy_from_kernel_nofault (char *dst,const void *unsafe_addr,long count);
@@ -324,7 +326,7 @@ long strnlen_user_nofault(const void __user *unsafe_addr, long count);* Returns 0 on success, or -EFAULT. */ #define probe_kernel_address (addr, retval) \ - probe_kernel_read(&retval, addr, sizeof(retval)) + copy_from_kernel_nofault(&retval, addr, sizeof(retval))#ifndef user_access_begin#define user_access_begin (ptr,len) access_ok(ptr, len)
kernel/debug/debug_core.c +3 -3
@@ -171,18 +171,18 @@ int __weak kgdb_arch_set_breakpoint(struct kgdb_bkpt *bpt) {int err; - err =probe_kernel_read (bpt->saved_instr, (char *)bpt->bpt_addr, + err =copy_from_kernel_nofault (bpt->saved_instr, (char *)bpt->bpt_addr,BREAK_INSTR_SIZE );if (err)return err; - err =probe_kernel_write ((char *)bpt->bpt_addr, + err =copy_to_kernel_nofault ((char *)bpt->bpt_addr, arch_kgdb_ops.gdb_bpt_instr,BREAK_INSTR_SIZE );return err; }int __weak kgdb_arch_remove_breakpoint (struct kgdb_bkpt *bpt) { -return probe_kernel_write ((char *)bpt->bpt_addr, +return copy_to_kernel_nofault ((char *)bpt->bpt_addr, (char *)bpt->saved_instr,BREAK_INSTR_SIZE ); }
kernel/debug/gdbstub.c +3 -3
@@ -247,7 +247,7 @@ char *kgdb_mem2hex(char *mem, char *buf, int count)*/ tmp = buf + count; - err =probe_kernel_read (tmp, mem, count); + err =copy_from_kernel_nofault (tmp, mem, count);if (err)return NULL ;while (count >0 ) {
@@ -283,7 +283,7 @@ int kgdb_hex2mem(char *buf, char *mem, int count) *tmp_raw |=hex_to_bin (*tmp_hex--) <<4 ; } -return probe_kernel_write (mem, tmp_raw, count); +return copy_to_kernel_nofault (mem, tmp_raw, count); }/*
@@ -335,7 +335,7 @@ static int kgdb_ebin2mem(char *buf, char *mem, int count) size++; } -return probe_kernel_write (mem, c, size); +return copy_to_kernel_nofault (mem, c, size); }#if DBG_MAX_REG_NUM >0
kernel/debug/kdb/kdb_main.c +2 -1
@@ -2325,7 +2325,8 @@ void kdb_ps1(const struct task_struct *p)int cpu;unsigned long tmp; -if (!p ||probe_kernel_read (&tmp, (char *)p,sizeof (unsigned long ))) +if (!p || +copy_from_kernel_nofault (&tmp, (char *)p,sizeof (unsigned long )))return ; cpu =kdb_process_cpu (p);
kernel/debug/kdb/kdb_support.c +4 -3
@@ -325,7 +325,7 @@ char *kdb_strdup(const char *str, gfp_t type)*/ int kdb_getarea_size (void *res,unsigned long addr,size_t size) { -int ret =probe_kernel_read ((char *)res, (char *)addr, size); +int ret =copy_from_kernel_nofault ((char *)res, (char *)addr, size);if (ret) {if (!KDB_STATE (SUPPRESS )) {kdb_printf ("kdb_getarea: Bad address 0x%lx\n" , addr);
@@ -350,7 +350,7 @@ int kdb_getarea_size(void *res, unsigned long addr, size_t size)*/ int kdb_putarea_size (unsigned long addr,void *res,size_t size) { -int ret =probe_kernel_read ((char *)addr, (char *)res, size); +int ret =copy_from_kernel_nofault ((char *)addr, (char *)res, size);if (ret) {if (!KDB_STATE (SUPPRESS )) {kdb_printf ("kdb_putarea: Bad address 0x%lx\n" , addr);
@@ -624,7 +624,8 @@ char kdb_task_state_char (const struct task_struct *p)char state;unsigned long tmp; -if (!p ||probe_kernel_read (&tmp, (char *)p,sizeof (unsigned long ))) +if (!p || +copy_from_kernel_nofault (&tmp, (char *)p,sizeof (unsigned long )))return 'E' ; cpu =kdb_process_cpu (p);
kernel/kthread.c +1 -1
@@ -179,7 +179,7 @@ void *kthread_probe_data(struct task_struct *task)struct kthread *kthread =to_kthread (task);void *data =NULL ; -probe_kernel_read (&data, &kthread->data,sizeof (data)); +copy_from_kernel_nofault (&data, &kthread->data,sizeof (data));return data; }
kernel/trace/bpf_trace.c +1 -1
@@ -186,7 +186,7 @@ bpf_probe_read_kernel_common(void *dst, u32 size, const void *unsafe_ptr,if (unlikely (ret <0 )) goto fail; - ret =probe_kernel_read (dst, unsafe_ptr, size); + ret =copy_from_kernel_nofault (dst, unsafe_ptr, size);if (unlikely (ret <0 )) {if (compat) ret =probe_user_read (dst, user_ptr, size);
kernel/trace/trace_kprobe.c +2 -2
@@ -1208,7 +1208,7 @@ fetch_store_strlen(unsigned long addr)u8 c; do { - ret = probe_kernel_read(&c, (u8 *)addr + len, 1); + ret = copy_from_kernel_nofault(&c, (u8 *)addr + len, 1); if (ret) ret = probe_user_read(&c, (__force u8 __user *)addr + len, 1);
@@ -1287,7 +1287,7 @@ probe_mem_read(void *dest, void *src, size_t size)const void __user *user_ptr = (__force const void __user *)src;int ret; - ret =probe_kernel_read (dest, src, size); + ret =copy_from_kernel_nofault (dest, src, size);if (ret) ret =probe_user_read (dest, user_ptr, size);return ret;
kernel/workqueue.c +5 -5
@@ -4627,11 +4627,11 @@ void print_worker_info(const char *log_lvl, struct task_struct *task)* Carefully copy the associated workqueue's workfn, name and desc. * Keep the original last '\0' in case the original is garbage. */ -probe_kernel_read (&fn, &worker->current_func,sizeof (fn)); -probe_kernel_read (&pwq, &worker->current_pwq,sizeof (pwq)); -probe_kernel_read (&wq, &pwq->wq,sizeof (wq)); -probe_kernel_read (name, wq->name,sizeof (name) -1 ); -probe_kernel_read (desc, worker->desc,sizeof (desc) -1 ); +copy_from_kernel_nofault (&fn, &worker->current_func,sizeof (fn)); +copy_from_kernel_nofault (&pwq, &worker->current_pwq,sizeof (pwq)); +copy_from_kernel_nofault (&wq, &pwq->wq,sizeof (wq)); +copy_from_kernel_nofault (name, wq->name,sizeof (name) -1 ); +copy_from_kernel_nofault (desc, worker->desc,sizeof (desc) -1 );if (fn || name[0 ] || desc[0 ]) {printk ("%sWorkqueue: %s %ps" , log_lvl, name, fn);
mm/maccess.c +24 -24
129 lines not on this page · raw message
mm/rodata_test.c +1 -1
@@ -25,7 +25,7 @@ void rodata_test(void) }/* test 2: write to the variable; this should fault */ -if (!probe_kernel_write ((void *)&rodata_test_data, +if (!copy_to_kernel_nofault ((void *)&rodata_test_data, (void *)&zero,sizeof (zero))) {pr_err ("test data was not read only\n" );return ;
mm/slub.c +1 -1
@@ -292,7 +292,7 @@ static inline void *get_freepointer_safe(struct kmem_cache *s, void *object)return get_freepointer (s, object); freepointer_addr = (unsigned long )object + s->offset; -probe_kernel_read (&p, (void **)freepointer_addr,sizeof (p)); +copy_from_kernel_nofault (&p, (void **)freepointer_addr,sizeof (p));return freelist_ptr (s, p, freepointer_addr); }
Signature · 7 lines
2.26.2 _______________________________________________ linux-um mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-um
PATCH 17/18 maccess: rename probe_user_{read,write} to copy_{from,to}_user_nofault
In reply to Christoph HellwigBetter describe what these functions do.
--- arch/powerpc/kernel/process.c | 3 ++- arch/powerpc/kvm/book3s_64_mmu_radix.c | 4 ++-- arch/powerpc/mm/fault.c | 2 +- arch/powerpc/oprofile/backtrace.c | 6 ++++-- arch/powerpc/perf/callchain_32.c | 2 +- arch/powerpc/perf/callchain_64.c | 2 +- arch/powerpc/perf/core-book3s.c | 3 ++- arch/powerpc/sysdev/fsl_pci.c | 4 ++-- include/linux/uaccess.h | 4 ++-- kernel/trace/bpf_trace.c | 6 +++--- kernel/trace/trace_kprobe.c | 6 +++--- mm/maccess.c | 10 +++++----- 12 files changed, 28 insertions(+), 24 deletions(-)
arch/powerpc/kernel/process.c +2 -1
@@ -1294,7 +1294,8 @@ void show_user_instructions(struct pt_regs *regs)for (i =0 ; i <8 && n; i++, n--, pc +=sizeof (int )) {int instr; -if (probe_user_read (&instr, (void __user *)pc,sizeof (instr))) { +if (copy_from_user_nofault (&instr, (void __user *)pc, +sizeof (instr))) {seq_buf_printf (&s,"XXXXXXXX " );continue ; }
arch/powerpc/kvm/book3s_64_mmu_radix.c +2 -2
@@ -64,9 +64,9 @@ unsigned long __kvmhv_copy_tofrom_guest_radix(int lpid, int pid,isync ();if (is_load) - ret =probe_user_read (to, (const void __user *)from, n); + ret =copy_from_user_nofault (to, (const void __user *)from, n);else - ret =probe_user_write ((void __user *)to, from, n); + ret =copy_to_user_nofault ((void __user *)to, from, n);/* switch the pid first to avoid running host with unallocated pid */ if (quadrant ==1 && pid != old_pid)
arch/powerpc/mm/fault.c +1 -1
@@ -280,7 +280,7 @@ static bool bad_stack_expansion(struct pt_regs *regs, unsigned long address, access_ok(nip, sizeof(*nip))) {unsigned int inst; -if (!probe_user_read (&inst, nip,sizeof (inst))) +if (!copy_from_user_nofault (&inst, nip,sizeof (inst)))return !store_updates_sp (inst); *must_retry = true; }
arch/powerpc/oprofile/backtrace.c +4 -2
@@ -33,7 +33,8 @@ static unsigned int user_getsp32(unsigned int sp, int is_first)* which means that we've done all that we can do from * interrupt context. */ -if (probe_user_read (stack_frame, (void __user *)p,sizeof (stack_frame))) +if (copy_from_user_nofault (stack_frame, (void __user *)p, +sizeof (stack_frame)))return 0 ;if (!is_first)
@@ -51,7 +52,8 @@ static unsigned long user_getsp64(unsigned long sp, int is_first) {unsigned long stack_frame[3 ]; -if (probe_user_read (stack_frame, (void __user *)sp,sizeof (stack_frame))) +if (copy_from_user_nofault (stack_frame, (void __user *)sp, +sizeof (stack_frame)))return 0 ;if (!is_first)
arch/powerpc/perf/callchain_32.c +1 -1
@@ -45,7 +45,7 @@ static int read_user_stack_32(unsigned int __user *ptr, unsigned int *ret) ((unsigned long)ptr & 3))return -EFAULT ; - rc =probe_user_read (ret, ptr,sizeof (*ret)); + rc =copy_from_user_nofault (ret, ptr,sizeof (*ret));if (IS_ENABLED (CONFIG_PPC64 ) && rc)return read_user_stack_slow (ptr, ret,4 );
arch/powerpc/perf/callchain_64.c +1 -1
@@ -71,7 +71,7 @@ static int read_user_stack_64(unsigned long __user *ptr, unsigned long *ret) ((unsigned long)ptr & 7))return -EFAULT ; -if (!probe_user_read (ret, ptr,sizeof (*ret))) +if (!copy_from_user_nofault (ret, ptr,sizeof (*ret)))return 0 ;return read_user_stack_slow (ptr, ret,8 );
arch/powerpc/perf/core-book3s.c +2 -1
@@ -426,7 +426,8 @@ static __u64 power_pmu_bhrb_to(u64 addr) }/* Userspace: need copy instruction here then translate it */ -if (probe_user_read (&instr, (unsigned int __user *)addr,sizeof (instr))) +if (copy_from_user_nofault (&instr, (unsigned int __user *)addr, +sizeof (instr)))return 0 ; target =branch_target (&instr);
arch/powerpc/sysdev/fsl_pci.c +2 -2
@@ -1066,8 +1066,8 @@ int fsl_pci_mcheck_exception(struct pt_regs *regs)if (is_in_pci_mem_space (addr)) {if (user_mode (regs)) - ret =probe_user_read (&inst, (void __user *)regs->nip, -sizeof (inst)); + ret =copy_from_user_nofault (&inst, + (void __user *)regs->nip,sizeof (inst));else ret =probe_kernel_address ((void *)regs->nip, inst);
include/linux/uaccess.h +2 -2
@@ -307,8 +307,8 @@ bool copy_from_kernel_nofault_allowed(void *dst, const void *unsafe_src,long copy_from_kernel_nofault (void *dst,const void *src,size_t size);long notrace copy_to_kernel_nofault (void *dst,const void *src,size_t size); -extern long probe_user_read (void *dst,const void __user *src,size_t size); -extern long notrace probe_user_write (void __user *dst,const void *src, +long copy_from_user_nofault (void *dst,const void __user *src,size_t size); +long notrace copy_to_user_nofault (void __user *dst,const void *src,size_t size); long strncpy_from_kernel_nofault(char *dst, const void *unsafe_addr,
kernel/trace/bpf_trace.c +3 -3
@@ -139,7 +139,7 @@ static const struct bpf_func_proto bpf_override_return_proto = { BPF_CALL_3(bpf_probe_read_user, void *, dst, u32, size, const void __user *, unsafe_ptr) { -int ret =probe_user_read (dst, unsafe_ptr, size); +int ret =copy_from_user_nofault (dst, unsafe_ptr, size);if (unlikely (ret <0 ))memset (dst,0 , size);
@@ -189,7 +189,7 @@ bpf_probe_read_kernel_common(void *dst, u32 size, const void *unsafe_ptr, ret =copy_from_kernel_nofault (dst, unsafe_ptr, size);if (unlikely (ret <0 )) {if (compat) - ret =probe_user_read (dst, user_ptr, size); + ret =copy_from_user_nofault (dst, user_ptr, size);if (unlikely (ret <0 )) goto fail; }
@@ -316,7 +316,7 @@ BPF_CALL_3(bpf_probe_write_user, void __user *, unsafe_ptr, const void *, src,if (unlikely (!nmi_uaccess_okay ()))return -EPERM ; -return probe_user_write (unsafe_ptr, src, size); +return copy_to_user_nofault (unsafe_ptr, src, size); }static const struct bpf_func_proto bpf_probe_write_user_proto = {
kernel/trace/trace_kprobe.c +3 -3
@@ -1210,7 +1210,7 @@ fetch_store_strlen(unsigned long addr)do { ret =copy_from_kernel_nofault (&c, (u8 *)addr + len,1 );if (ret) - ret =probe_user_read (&c, + ret =copy_from_user_nofault (&c, (__force u8 __user *)addr + len,1 ); len++; }while (c && ret ==0 && len <MAX_STRING_SIZE );
@@ -1289,7 +1289,7 @@ probe_mem_read(void *dest, void *src, size_t size) ret =copy_from_kernel_nofault (dest, src, size);if (ret) - ret =probe_user_read (dest, user_ptr, size); + ret =copy_from_user_nofault (dest, user_ptr, size);return ret; }
@@ -1298,7 +1298,7 @@ probe_mem_read_user(void *dest, void *src, size_t size) {const void __user *uaddr = (__force const void __user *)src; -return probe_user_read (dest, uaddr, size); +return copy_from_user_nofault (dest, uaddr, size); }/* Note that we don't verify it, since the code does not come from user space */
mm/maccess.c +5 -5
@@ -192,7 +192,7 @@ long strncpy_from_kernel_nofault(char *dst, const void *unsafe_addr, long count)#endif /* HAVE_ARCH_PROBE_KERNEL */ /** -* probe_user_read(): safely attempt to read from a user-space location +* copy_from_user_nofault(): safely attempt to read from a user-space location * @dst: pointer to the buffer that shall take the data * @src: address to read from. This must be a user address. * @size: size of the data chunk
@@ -200,7 +200,7 @@ long strncpy_from_kernel_nofault(char *dst, const void *unsafe_addr, long count)* Safely read from user address @src to the buffer at @dst. If a kernel fault * happens, handle that and return -EFAULT. */ -long probe_user_read (void *dst,const void __user *src,size_t size) +long copy_from_user_nofault (void *dst,const void __user *src,size_t size) {long ret = -EFAULT ;mm_segment_t old_fs =get_fs ();
@@ -217,10 +217,10 @@ long probe_user_read(void *dst, const void __user *src, size_t size)return -EFAULT ;return 0 ; } -EXPORT_SYMBOL_GPL (probe_user_read); +EXPORT_SYMBOL_GPL (copy_from_user_nofault);/** -* probe_user_write(): safely attempt to write to a user-space location +* copy_to_user_nofault(): safely attempt to write to a user-space location * @dst: address to write to * @src: pointer to the data that shall be written * @size: size of the data chunk
@@ -228,7 +228,7 @@ EXPORT_SYMBOL_GPL(probe_user_read);* Safely write to address @dst from the buffer at @src. If a kernel fault * happens, handle that and return -EFAULT. */ -long probe_user_write (void __user *dst,const void *src,size_t size) +long copy_to_user_nofault (void __user *dst,const void *src,size_t size) {long ret = -EFAULT ;mm_segment_t old_fs =get_fs ();
Signature · 7 lines
2.26.2 _______________________________________________ linux-um mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-um
PATCH 18/18 maccess: rename probe_kernel_address to get_kernel_nofault
In reply to Christoph HellwigBetter describe what this helper does, and match the naming of copy_from_kernel_nofault.
--- arch/arm/kernel/traps.c | 2 +- arch/arm/mm/alignment.c | 4 ++-- arch/arm64/kernel/traps.c | 2 +- arch/ia64/include/asm/sections.h | 2 +- arch/parisc/kernel/process.c | 2 +- arch/powerpc/include/asm/sections.h | 2 +- arch/powerpc/kernel/kgdb.c | 2 +- arch/powerpc/kernel/process.c | 2 +- arch/powerpc/sysdev/fsl_pci.c | 2 +- arch/riscv/kernel/traps.c | 4 ++-- arch/s390/mm/fault.c | 2 +- arch/sh/kernel/traps.c | 2 +- arch/x86/kernel/probe_roms.c | 20 ++++++++++---------- arch/x86/kernel/traps.c | 2 +- arch/x86/mm/fault.c | 6 +++--- arch/x86/pci/pcbios.c | 2 +- include/linux/uaccess.h | 4 ++-- lib/test_lockup.c | 6 +++--- 18 files changed, 34 insertions(+), 34 deletions(-)
arch/arm/kernel/traps.c +1 -1
@@ -391,7 +391,7 @@ int is_valid_bugaddr(unsigned long pc)u32 insn =__opcode_to_mem_arm (BUG_INSTR_VALUE );#endif -if (probe_kernel_address ((unsigned *)pc, bkpt)) +if (get_kernel_nofault ((unsigned *)pc, bkpt))return 0 ;return bkpt == insn;
arch/arm/mm/alignment.c +2 -2
@@ -774,7 +774,7 @@ static int alignment_get_arm(struct pt_regs *regs, u32 *ip, u32 *inst)if (user_mode (regs)) fault =get_user (instr, ip);else - fault =probe_kernel_address (ip, instr); + fault =get_kernel_nofault (ip, instr); *inst =__mem_to_opcode_arm (instr);
@@ -789,7 +789,7 @@ static int alignment_get_thumb(struct pt_regs *regs, u16 *ip, u16 *inst)if (user_mode (regs)) fault =get_user (instr, ip);else - fault =probe_kernel_address (ip, instr); + fault =get_kernel_nofault (ip, instr); *inst =__mem_to_opcode_thumb16 (instr);
arch/arm64/kernel/traps.c +1 -1
@@ -315,7 +315,7 @@ static int call_undef_hook(struct pt_regs *regs)if (!user_mode (regs)) {__le32 instr_le; -if (probe_kernel_address ((__force __le32 *)pc, instr_le)) +if (get_kernel_nofault ((__force __le32 *)pc, instr_le)) goto exit; instr =le32_to_cpu (instr_le); }else if (compat_thumb_mode (regs)) {
arch/ia64/include/asm/sections.h +1 -1
@@ -35,7 +35,7 @@ static inline void *dereference_function_descriptor(void *ptr)struct fdesc *desc = ptr;void *p; -if (!probe_kernel_address (&desc->ip, p)) +if (!get_kernel_nofault (&desc->ip, p)) ptr = p;return ptr; }
arch/parisc/kernel/process.c +1 -1
@@ -293,7 +293,7 @@ void *dereference_function_descriptor(void *ptr)Elf64_Fdesc *desc = ptr;void *p; -if (!probe_kernel_address (&desc->addr, p)) +if (!get_kernel_nofault (&desc->addr, p)) ptr = p;return ptr; }
arch/powerpc/include/asm/sections.h +1 -1
@@ -85,7 +85,7 @@ static inline void *dereference_function_descriptor(void *ptr)struct ppc64_opd_entry *desc = ptr;void *p; -if (!probe_kernel_address (&desc->funcaddr, p)) +if (!get_kernel_nofault (&desc->funcaddr, p)) ptr = p;return ptr; }
arch/powerpc/kernel/kgdb.c +1 -1
@@ -420,7 +420,7 @@ int kgdb_arch_set_breakpoint(struct kgdb_bkpt *bpt)unsigned int instr;unsigned int *addr = (unsigned int *)bpt->bpt_addr; - err =probe_kernel_address (addr, instr); + err =get_kernel_nofault (addr, instr);if (err)return err;
arch/powerpc/kernel/process.c +1 -1
@@ -1260,7 +1260,7 @@ static void show_instructions(struct pt_regs *regs)#endif if (!__kernel_text_address (pc) || -probe_kernel_address ((const void *)pc, instr)) { +get_kernel_nofault ((const void *)pc, instr)) {pr_cont ("XXXXXXXX " ); }else {if (regs->nip == pc)
arch/powerpc/sysdev/fsl_pci.c +1 -1
@@ -1069,7 +1069,7 @@ int fsl_pci_mcheck_exception(struct pt_regs *regs) ret =copy_from_user_nofault (&inst, (void __user *)regs->nip,sizeof (inst)); else - ret =probe_kernel_address ((void *)regs->nip, inst); + ret =get_kernel_nofault ((void *)regs->nip, inst);if (!ret &&mcheck_handle_load (regs, inst)) { regs->nip +=4 ;
arch/riscv/kernel/traps.c +2 -2
@@ -137,7 +137,7 @@ static inline unsigned long get_break_insn_length(unsigned long pc) {bug_insn_t insn; -if (probe_kernel_address ((bug_insn_t *)pc, insn)) +if (get_kernel_nofault ((bug_insn_t *)pc, insn))return 0 ;return GET_INSN_LENGTH (insn);
@@ -160,7 +160,7 @@ int is_valid_bugaddr(unsigned long pc)if (pc <VMALLOC_START )return 0 ; -if (probe_kernel_address ((bug_insn_t *)pc, insn)) +if (get_kernel_nofault ((bug_insn_t *)pc, insn))return 0 ;if ((insn & __INSN_LENGTH_MASK) == __INSN_LENGTH_32)return (insn == __BUG_INSN_32);
arch/s390/mm/fault.c +1 -1
@@ -106,7 +106,7 @@ static int bad_address(void *p) {unsigned long dummy; -return probe_kernel_address ((unsigned long *)p, dummy); +return get_kernel_nofault ((unsigned long *)p, dummy); }static void dump_pagetable (unsigned long asce,unsigned long address)
arch/sh/kernel/traps.c +1 -1
@@ -118,7 +118,7 @@ int is_valid_bugaddr(unsigned long addr)if (addr <PAGE_OFFSET )return 0 ; -if (probe_kernel_address ((insn_size_t *)addr, opcode)) +if (get_kernel_nofault ((insn_size_t *)addr, opcode))return 0 ;if (opcode ==TRAPA_BUG_OPCODE )return 1 ;
arch/x86/kernel/probe_roms.c +10 -10
@@ -99,7 +99,7 @@ static bool probe_list(struct pci_dev *pdev, unsigned short vendor,unsigned short device; do { - if (probe_kernel_address(rom_list, device) != 0) + if (get_kernel_nofault(rom_list, device) != 0) device = 0; if (device && match_id(pdev, vendor, device))
@@ -125,13 +125,13 @@ static struct resource *find_oprom(struct pci_dev *pdev)break ; rom =isa_bus_to_virt (res->start); -if (probe_kernel_address (rom +0x18 , offset) !=0 ) +if (get_kernel_nofault (rom +0x18 , offset) !=0 )continue ; -if (probe_kernel_address (rom + offset +0x4 , vendor) !=0 ) +if (get_kernel_nofault (rom + offset +0x4 , vendor) !=0 )continue ; -if (probe_kernel_address (rom + offset +0x6 , device) !=0 ) +if (get_kernel_nofault (rom + offset +0x6 , device) !=0 )continue ;if (match_id (pdev, vendor, device)) {
@@ -139,8 +139,8 @@ static struct resource *find_oprom(struct pci_dev *pdev)break ; } -if (probe_kernel_address (rom + offset +0x8 , list) ==0 && -probe_kernel_address (rom + offset +0xc , rev) ==0 && +if (get_kernel_nofault (rom + offset +0x8 , list) ==0 && +get_kernel_nofault (rom + offset +0xc , rev) ==0 && rev >=3 && list &&probe_list (pdev, vendor, rom + offset + list)) { oprom = res;
@@ -183,14 +183,14 @@ static int __init romsignature(const unsigned char *rom)const unsigned short *const ptr = (const unsigned short *)rom;unsigned short sig; -return probe_kernel_address (ptr, sig) ==0 && sig ==ROMSIGNATURE ; +return get_kernel_nofault (ptr, sig) ==0 && sig ==ROMSIGNATURE ; }static int __init romchecksum (const unsigned char *rom,unsigned long length) {unsigned char sum, c; -for (sum =0 ; length &&probe_kernel_address (rom++, c) ==0 ; length--) +for (sum =0 ; length &&get_kernel_nofault (rom++, c) ==0 ; length--) sum += c;return !length && !sum; }
@@ -211,7 +211,7 @@ void __init probe_roms(void) video_rom_resource.start = start; -if (probe_kernel_address (rom +2 , c) !=0 ) +if (get_kernel_nofault (rom +2 , c) !=0 )continue ;/* 0 < length <= 0x7f * 512, historically */
@@ -249,7 +249,7 @@ void __init probe_roms(void)if (!romsignature (rom))continue ; -if (probe_kernel_address (rom +2 , c) !=0 ) +if (get_kernel_nofault (rom +2 , c) !=0 )continue ;/* 0 < length <= 0x7f * 512, historically */
arch/x86/kernel/traps.c +1 -1
@@ -161,7 +161,7 @@ int is_valid_bugaddr(unsigned long addr)if (addr <TASK_SIZE_MAX )return 0 ; -if (probe_kernel_address ((unsigned short *)addr, ud)) +if (get_kernel_nofault ((unsigned short *)addr, ud))return 0 ;return ud ==INSN_UD0 || ud ==INSN_UD2 ;
arch/x86/mm/fault.c +3 -3
@@ -98,7 +98,7 @@ check_prefetch_opcode(struct pt_regs *regs, unsigned char *instr,return !instr_lo || (instr_lo>>1 ) ==1 ;case 0x00 :/* Prefetch instruction is 0x0F0D or 0x0F18 */ -if (probe_kernel_address (instr, opcode)) +if (get_kernel_nofault (instr, opcode))return 0 ; *prefetch = (instr_lo == 0xF) &&
@@ -132,7 +132,7 @@ is_prefetch(struct pt_regs *regs, unsigned long error_code, unsigned long addr)while (instr < max_instr) {unsigned char opcode; -if (probe_kernel_address (instr, opcode)) +if (get_kernel_nofault (instr, opcode))break ; instr++;
@@ -441,7 +441,7 @@ static int bad_address(void *p) {unsigned long dummy; -return probe_kernel_address ((unsigned long *)p, dummy); +return get_kernel_nofault ((unsigned long *)p, dummy); }static void dump_pagetable (unsigned long address)
arch/x86/pci/pcbios.c +1 -1
@@ -302,7 +302,7 @@ static const struct pci_raw_ops *__init pci_find_bios(void) check <= (union bios32 *)__va (0xffff0 ); ++check) {long sig; -if (probe_kernel_address (&check->fields.signature, sig)) +if (get_kernel_nofault (&check->fields.signature, sig))continue ;if (check->fields.signature !=BIOS32_SIGNATURE )
include/linux/uaccess.h +2 -2
@@ -319,13 +319,13 @@ long strncpy_from_user_nofault(char *dst, const void __user *unsafe_addr,long strnlen_user_nofault (const void __user *unsafe_addr,long count);/** -* probe_kernel_address(): safely attempt to read from a location +* get_kernel_nofault(): safely attempt to read from a location * @addr: address to read from * @retval: read into this variable * * Returns 0 on success, or -EFAULT. */ -#define probe_kernel_address (addr, retval) \ +#define get_kernel_nofault (addr, retval) \ copy_from_kernel_nofault(&retval, addr, sizeof(retval))#ifndef user_access_begin
lib/test_lockup.c +3 -3
@@ -419,8 +419,8 @@ static bool test_kernel_ptr(unsigned long addr, int size)/* should be at least readable kernel address */ if (access_ok (ptr,1 ) ||access_ok (ptr + size -1 ,1 ) || -probe_kernel_address (ptr, buf) || -probe_kernel_address (ptr + size -1 , buf)) { +get_kernel_nofault (ptr, buf) || +get_kernel_nofault (ptr + size -1 , buf)) {pr_err ("invalid kernel ptr: %#lx\n" , addr);return true; }
@@ -437,7 +437,7 @@ static bool __maybe_unused test_magic(unsigned long addr, int offset,if (!addr)return false; -if (probe_kernel_address (ptr, magic) || magic != expected) { +if (get_kernel_nofault (ptr, magic) || magic != expected) {pr_err ("invalid magic at %#lx + %#x = %#x, expected %#x\n" , addr, offset, magic, expected);return true;
Signature · 7 lines
2.26.2 _______________________________________________ linux-um mailing list [email protected] http://lists.infradead.org/mailman/listinfo/linux-um
Quoted 6 lines · Christoph Hellwig
this series start cleaning up the safe kernel and user memory probing helpers in
mm/maccess.c, and then allows architectures to implement the kernel probing without overriding the address space limit and temporarily allowing access to user memory. It then switches x86 over to this new mechanism by reusing the unsafe_* uaccess logic.
Ok, I think I found a bug, and I had one more suggestion, but other than the two emails I sent this all looks like an improvement to me.
Linus
Daniel Borkmann
maintainerQuoted 8 lines · Christoph Hellwig
Hi all,
this series start cleaning up the safe kernel and user memory probing helpers in
mm/maccess.c, and then allows architectures to implement the kernel probing without overriding the address space limit and temporarily allowing access to user memory. It then switches x86 over to this new mechanism by reusing the unsafe_* uaccess logic.
Quoted 6 lines
This version also switches to the saner copy_{from,to}_kernel_nofault naming suggested by Linus.
I kept the x86 helprs as-is without calling unsage_{get,put}_user as that avoids a number of hard to trace casts, and it will still work with the asm-goto based version easily.
Aside from comments on list, the series looks reasonable to me. For BPF the bpf_probe_read() helper would be slightly penalized for probing user memory given we now test on copy_from_kernel_nofault() first and if that fails only then fall back to copy_from_user_nofault(), but it seems small enough that it shouldn't matter too much and aside from that we have the newer bpf_probe_read_kernel() and bpf_probe_read_user() anyway that BPF progs should use instead, so I think it's okay.
For patch 14 and patch 15, do you roughly know the performance gain with the new probe_kernel_read_loop() + arch_kernel_read() approach?
Thanks,
Daniel
Quoted 5 lines · Daniel Borkmann
Aside from comments on list, the series looks reasonable to me. For BPF the
bpf_probe_read()helper would be slightly penalized for probing user memory given we now test oncopy_from_kernel_nofault()first and if that fails only then fall back tocopy_from_user_nofault(),
Again, no.
If you can't tell that one or the other is always the right thing, then that function is simply buggy and wrong.
On sparc and on s390, address X can be _both_ a kernel address and a user address. You need to specify which it is (by using the proper function). The whole "try one first, then the other" doesn't work. They may both "work", and by virtue of that, unless you can state "yes, we always want user space" or "yes, we always want kernel", that "try one or the other" isn't valid.
And it can be a real security issue. If a user program can be made to read kernel memory when BPF validated things as a user pointer, it's an obvious security issue.
But it can be a security issue the other way around too: if the BPF code expects to get a kernel string, but user space can fool it into reading a user string instead by mapping something of its own into the user space address that aliases the kernel space address, then you can presumably fool the BPF program to do bad things too (eg mess up any BPF packet switching routines?).
So BPF really really really needs to specify which one it is. Not specifying it and saying "whichever" is a bug, and a security issue.
Linus
Quoted 10 lines · Daniel Borkmann
Aside from comments on list, the series looks reasonable to me. For BPF the
bpf_probe_read()helper would be slightly penalized for probing user memory given we now test oncopy_from_kernel_nofault()first and if that fails only then fall back tocopy_from_user_nofault(), but it seems small enough that it shouldn't matter too much and aside from that we have the newerbpf_probe_read_kernel()andbpf_probe_read_user()anyway that BPF progs should use instead, so I think it's okay.For patch 14 and patch 15, do you roughly know the performance gain with the new
probe_kernel_read_loop()+arch_kernel_read()approach?
I don't think there should be any measurable difference in performance for typical use cases. We'll save the stac/clac pair, but that's it. The real eason is to avoid that stac/clac pair that opens up a window for explots, and as a significant enabler for killing of set_fs based address limit overrides entirely.
_______________________________________________
linux-um mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-um