Detect pointer math wrap-around (overflow and underflow) #344
Comments
3 tasks
kees
changed the title
|
As for bullet point 2: demo: https://godbolt.org/z/Ef3Kvqq1x
Update: I triaged and nikic fixed here: llvm/llvm-project#67772 |
sirlucjan
pushed a commit
to CachyOS/linux
that referenced
this issue
Jan 23, 2024
In an effort to separate intentional arithmetic wrap-around from unexpected wrap-around, we need to refactor places that depend on this kind of math. One of the most common code patterns of this is: VAR + value < VAR Notably, this is considered "undefined behavior" for signed and pointer types, which the kernel works around by using the -fno-strict-overflow option in the build[1] (which used to just be -fwrapv). Regardless, we want to get the kernel source to the position where we can meaningfully instrument arithmetic wrap-around conditions and catch them when they are unexpected, regardless of whether they are signed[2], unsigned[3], or pointer[4] types. Switch to a more regular type for a 64-bit value and refactor the open-coded wrap-around addition test to use subtraction from the type max (since add_would_overflow() may not be defined in early boot code). This paves the way to enabling the wrap-around sanitizers in the future. Link: https://git.kernel.org/linus/68df3755e383e6fecf2354a67b08f92f18536594 [1] Link: KSPP#26 [2] Link: KSPP#27 [3] Link: KSPP#344 [4] Cc: Nick Terrell <terrelln@fb.com> Cc: Paul Jones <paul@pauljones.id.au> Cc: Sedat Dilek <sedat.dilek@gmail.com> Cc: Oleksandr Natalenko <oleksandr@natalenko.name> Cc: Xin Gao <gaoxin@cdjrlc.com> Signed-off-by: Kees Cook <keescook@chromium.org>
sirlucjan
pushed a commit
to CachyOS/linux
that referenced
this issue
Jan 23, 2024
In an effort to separate intentional arithmetic wrap-around from unexpected wrap-around, we need to refactor places that depend on this kind of math. One of the most common code patterns of this is: VAR + value < VAR Notably, this is considered "undefined behavior" for signed and pointer types, which the kernel works around by using the -fno-strict-overflow option in the build[1] (which used to just be -fwrapv). Regardless, we want to get the kernel source to the position where we can meaningfully instrument arithmetic wrap-around conditions and catch them when they are unexpected, regardless of whether they are signed[2], unsigned[3], or pointer[4] types. Switch to a more regular type for a 64-bit value and refactor the open-coded wrap-around addition test to use subtraction from the type max (since add_would_overflow() may not be defined in early boot code). This paves the way to enabling the wrap-around sanitizers in the future. Link: https://git.kernel.org/linus/68df3755e383e6fecf2354a67b08f92f18536594 [1] Link: KSPP#26 [2] Link: KSPP#27 [3] Link: KSPP#344 [4] Cc: Nick Terrell <terrelln@fb.com> Cc: Paul Jones <paul@pauljones.id.au> Cc: Sedat Dilek <sedat.dilek@gmail.com> Cc: Oleksandr Natalenko <oleksandr@natalenko.name> Cc: Xin Gao <gaoxin@cdjrlc.com> Signed-off-by: Kees Cook <keescook@chromium.org>
sirlucjan
pushed a commit
to CachyOS/linux
that referenced
this issue
Jan 23, 2024
In an effort to separate intentional arithmetic wrap-around from unexpected wrap-around, we need to refactor places that depend on this kind of math. One of the most common code patterns of this is: VAR + value < VAR Notably, this is considered "undefined behavior" for signed and pointer types, which the kernel works around by using the -fno-strict-overflow option in the build[1] (which used to just be -fwrapv). Regardless, we want to get the kernel source to the position where we can meaningfully instrument arithmetic wrap-around conditions and catch them when they are unexpected, regardless of whether they are signed[2], unsigned[3], or pointer[4] types. Switch to a more regular type for a 64-bit value and refactor the open-coded wrap-around addition test to use subtraction from the type max (since add_would_overflow() may not be defined in early boot code). This paves the way to enabling the wrap-around sanitizers in the future. Link: https://git.kernel.org/linus/68df3755e383e6fecf2354a67b08f92f18536594 [1] Link: KSPP#26 [2] Link: KSPP#27 [3] Link: KSPP#344 [4] Cc: Nick Terrell <terrelln@fb.com> Cc: Paul Jones <paul@pauljones.id.au> Cc: Sedat Dilek <sedat.dilek@gmail.com> Cc: Oleksandr Natalenko <oleksandr@natalenko.name> Cc: Xin Gao <gaoxin@cdjrlc.com> Signed-off-by: Kees Cook <keescook@chromium.org>
lseman
pushed a commit
to CachyOS/linux
that referenced
this issue
Feb 8, 2024
In an effort to separate intentional arithmetic wrap-around from unexpected wrap-around, we need to refactor places that depend on this kind of math. One of the most common code patterns of this is: VAR + value < VAR Notably, this is considered "undefined behavior" for signed and pointer types, which the kernel works around by using the -fno-strict-overflow option in the build[1] (which used to just be -fwrapv). Regardless, we want to get the kernel source to the position where we can meaningfully instrument arithmetic wrap-around conditions and catch them when they are unexpected, regardless of whether they are signed[2], unsigned[3], or pointer[4] types. Switch to a more regular type for a 64-bit value and refactor the open-coded wrap-around addition test to use subtraction from the type max (since add_would_overflow() may not be defined in early boot code). This paves the way to enabling the wrap-around sanitizers in the future. Link: https://git.kernel.org/linus/68df3755e383e6fecf2354a67b08f92f18536594 [1] Link: KSPP#26 [2] Link: KSPP#27 [3] Link: KSPP#344 [4] Cc: Nick Terrell <terrelln@fb.com> Cc: Paul Jones <paul@pauljones.id.au> Cc: Sedat Dilek <sedat.dilek@gmail.com> Cc: Oleksandr Natalenko <oleksandr@natalenko.name> Cc: Xin Gao <gaoxin@cdjrlc.com> Signed-off-by: Kees Cook <keescook@chromium.org>
ptr1337
pushed a commit
to CachyOS/linux
that referenced
this issue
Mar 25, 2024
In an effort to separate intentional arithmetic wrap-around from unexpected wrap-around, we need to refactor places that depend on this kind of math. One of the most common code patterns of this is: VAR + value < VAR Notably, this is considered "undefined behavior" for signed and pointer types, which the kernel works around by using the -fno-strict-overflow option in the build[1] (which used to just be -fwrapv). Regardless, we want to get the kernel source to the position where we can meaningfully instrument arithmetic wrap-around conditions and catch them when they are unexpected, regardless of whether they are signed[2], unsigned[3], or pointer[4] types. Switch to a more regular type for a 64-bit value and refactor the open-coded wrap-around addition test to use subtraction from the type max (since add_would_overflow() may not be defined in early boot code). This paves the way to enabling the wrap-around sanitizers in the future. Link: https://git.kernel.org/linus/68df3755e383e6fecf2354a67b08f92f18536594 [1] Link: KSPP#26 [2] Link: KSPP#27 [3] Link: KSPP#344 [4] Cc: Nick Terrell <terrelln@fb.com> Cc: Paul Jones <paul@pauljones.id.au> Cc: Sedat Dilek <sedat.dilek@gmail.com> Cc: Oleksandr Natalenko <oleksandr@natalenko.name> Cc: Xin Gao <gaoxin@cdjrlc.com> Signed-off-by: Kees Cook <keescook@chromium.org>
sirlucjan
pushed a commit
to CachyOS/linux
that referenced
this issue
Mar 27, 2024
In an effort to separate intentional arithmetic wrap-around from unexpected wrap-around, we need to refactor places that depend on this kind of math. One of the most common code patterns of this is: VAR + value < VAR Notably, this is considered "undefined behavior" for signed and pointer types, which the kernel works around by using the -fno-strict-overflow option in the build[1] (which used to just be -fwrapv). Regardless, we want to get the kernel source to the position where we can meaningfully instrument arithmetic wrap-around conditions and catch them when they are unexpected, regardless of whether they are signed[2], unsigned[3], or pointer[4] types. Switch to a more regular type for a 64-bit value and refactor the open-coded wrap-around addition test to use subtraction from the type max (since add_would_overflow() may not be defined in early boot code). This paves the way to enabling the wrap-around sanitizers in the future. Link: https://git.kernel.org/linus/68df3755e383e6fecf2354a67b08f92f18536594 [1] Link: KSPP#26 [2] Link: KSPP#27 [3] Link: KSPP#344 [4] Cc: Nick Terrell <terrelln@fb.com> Cc: Paul Jones <paul@pauljones.id.au> Cc: Sedat Dilek <sedat.dilek@gmail.com> Cc: Oleksandr Natalenko <oleksandr@natalenko.name> Cc: Xin Gao <gaoxin@cdjrlc.com> Signed-off-by: Kees Cook <keescook@chromium.org>
sirlucjan
pushed a commit
to CachyOS/linux
that referenced
this issue
Mar 27, 2024
In an effort to separate intentional arithmetic wrap-around from unexpected wrap-around, we need to refactor places that depend on this kind of math. One of the most common code patterns of this is: VAR + value < VAR Notably, this is considered "undefined behavior" for signed and pointer types, which the kernel works around by using the -fno-strict-overflow option in the build[1] (which used to just be -fwrapv). Regardless, we want to get the kernel source to the position where we can meaningfully instrument arithmetic wrap-around conditions and catch them when they are unexpected, regardless of whether they are signed[2], unsigned[3], or pointer[4] types. Switch to a more regular type for a 64-bit value and refactor the open-coded wrap-around addition test to use subtraction from the type max (since add_would_overflow() may not be defined in early boot code). This paves the way to enabling the wrap-around sanitizers in the future. Link: https://git.kernel.org/linus/68df3755e383e6fecf2354a67b08f92f18536594 [1] Link: KSPP#26 [2] Link: KSPP#27 [3] Link: KSPP#344 [4] Cc: Nick Terrell <terrelln@fb.com> Cc: Paul Jones <paul@pauljones.id.au> Cc: Sedat Dilek <sedat.dilek@gmail.com> Cc: Oleksandr Natalenko <oleksandr@natalenko.name> Cc: Xin Gao <gaoxin@cdjrlc.com> Signed-off-by: Kees Cook <keescook@chromium.org>
sirlucjan
pushed a commit
to CachyOS/linux
that referenced
this issue
Mar 27, 2024
In an effort to separate intentional arithmetic wrap-around from unexpected wrap-around, we need to refactor places that depend on this kind of math. One of the most common code patterns of this is: VAR + value < VAR Notably, this is considered "undefined behavior" for signed and pointer types, which the kernel works around by using the -fno-strict-overflow option in the build[1] (which used to just be -fwrapv). Regardless, we want to get the kernel source to the position where we can meaningfully instrument arithmetic wrap-around conditions and catch them when they are unexpected, regardless of whether they are signed[2], unsigned[3], or pointer[4] types. Switch to a more regular type for a 64-bit value and refactor the open-coded wrap-around addition test to use subtraction from the type max (since add_would_overflow() may not be defined in early boot code). This paves the way to enabling the wrap-around sanitizers in the future. Link: https://git.kernel.org/linus/68df3755e383e6fecf2354a67b08f92f18536594 [1] Link: KSPP#26 [2] Link: KSPP#27 [3] Link: KSPP#344 [4] Cc: Nick Terrell <terrelln@fb.com> Cc: Paul Jones <paul@pauljones.id.au> Cc: Sedat Dilek <sedat.dilek@gmail.com> Cc: Oleksandr Natalenko <oleksandr@natalenko.name> Cc: Xin Gao <gaoxin@cdjrlc.com> Signed-off-by: Kees Cook <keescook@chromium.org>
nilz3000
pushed a commit
to nilz3000/pf-linux
that referenced
this issue
Apr 10, 2024
In an effort to separate intentional arithmetic wrap-around from unexpected wrap-around, we need to refactor places that depend on this kind of math. One of the most common code patterns of this is: VAR + value < VAR Notably, this is considered "undefined behavior" for signed and pointer types, which the kernel works around by using the -fno-strict-overflow option in the build[1] (which used to just be -fwrapv). Regardless, we want to get the kernel source to the position where we can meaningfully instrument arithmetic wrap-around conditions and catch them when they are unexpected, regardless of whether they are signed[2], unsigned[3], or pointer[4] types. Switch to a more regular type for a 64-bit value and refactor the open-coded wrap-around addition test to use subtraction from the type max (since add_would_overflow() may not be defined in early boot code). This paves the way to enabling the wrap-around sanitizers in the future. Link: https://git.kernel.org/linus/68df3755e383e6fecf2354a67b08f92f18536594 [1] Link: KSPP/linux#26 [2] Link: KSPP/linux#27 [3] Link: KSPP/linux#344 [4] Cc: Nick Terrell <terrelln@fb.com> Cc: Paul Jones <paul@pauljones.id.au> Cc: Sedat Dilek <sedat.dilek@gmail.com> Cc: Oleksandr Natalenko <oleksandr@natalenko.name> Cc: Xin Gao <gaoxin@cdjrlc.com> Signed-off-by: Kees Cook <keescook@chromium.org>
nilz3000
pushed a commit
to nilz3000/pf-linux
that referenced
this issue
Apr 14, 2024
In an effort to separate intentional arithmetic wrap-around from unexpected wrap-around, we need to refactor places that depend on this kind of math. One of the most common code patterns of this is: VAR + value < VAR Notably, this is considered "undefined behavior" for signed and pointer types, which the kernel works around by using the -fno-strict-overflow option in the build[1] (which used to just be -fwrapv). Regardless, we want to get the kernel source to the position where we can meaningfully instrument arithmetic wrap-around conditions and catch them when they are unexpected, regardless of whether they are signed[2], unsigned[3], or pointer[4] types. Switch to a more regular type for a 64-bit value and refactor the open-coded wrap-around addition test to use subtraction from the type max (since add_would_overflow() may not be defined in early boot code). This paves the way to enabling the wrap-around sanitizers in the future. Link: https://git.kernel.org/linus/68df3755e383e6fecf2354a67b08f92f18536594 [1] Link: KSPP/linux#26 [2] Link: KSPP/linux#27 [3] Link: KSPP/linux#344 [4] Cc: Nick Terrell <terrelln@fb.com> Cc: Paul Jones <paul@pauljones.id.au> Cc: Sedat Dilek <sedat.dilek@gmail.com> Cc: Oleksandr Natalenko <oleksandr@natalenko.name> Cc: Xin Gao <gaoxin@cdjrlc.com> Signed-off-by: Kees Cook <keescook@chromium.org>
ptr1337
pushed a commit
to CachyOS/linux
that referenced
this issue
May 13, 2024
In an effort to separate intentional arithmetic wrap-around from unexpected wrap-around, we need to refactor places that depend on this kind of math. One of the most common code patterns of this is: VAR + value < VAR Notably, this is considered "undefined behavior" for signed and pointer types, which the kernel works around by using the -fno-strict-overflow option in the build[1] (which used to just be -fwrapv). Regardless, we want to get the kernel source to the position where we can meaningfully instrument arithmetic wrap-around conditions and catch them when they are unexpected, regardless of whether they are signed[2], unsigned[3], or pointer[4] types. Switch to a more regular type for a 64-bit value and refactor the open-coded wrap-around addition test to use subtraction from the type max (since add_would_overflow() may not be defined in early boot code). This paves the way to enabling the wrap-around sanitizers in the future. Link: https://git.kernel.org/linus/68df3755e383e6fecf2354a67b08f92f18536594 [1] Link: KSPP#26 [2] Link: KSPP#27 [3] Link: KSPP#344 [4] Cc: Nick Terrell <terrelln@fb.com> Cc: Paul Jones <paul@pauljones.id.au> Cc: Sedat Dilek <sedat.dilek@gmail.com> Cc: Oleksandr Natalenko <oleksandr@natalenko.name> Cc: Xin Gao <gaoxin@cdjrlc.com> Signed-off-by: Kees Cook <keescook@chromium.org>
ptr1337
pushed a commit
to CachyOS/linux
that referenced
this issue
May 26, 2024
In an effort to separate intentional arithmetic wrap-around from unexpected wrap-around, we need to refactor places that depend on this kind of math. One of the most common code patterns of this is: VAR + value < VAR Notably, this is considered "undefined behavior" for signed and pointer types, which the kernel works around by using the -fno-strict-overflow option in the build[1] (which used to just be -fwrapv). Regardless, we want to get the kernel source to the position where we can meaningfully instrument arithmetic wrap-around conditions and catch them when they are unexpected, regardless of whether they are signed[2], unsigned[3], or pointer[4] types. Switch to a more regular type for a 64-bit value and refactor the open-coded wrap-around addition test to use subtraction from the type max (since add_would_overflow() may not be defined in early boot code). This paves the way to enabling the wrap-around sanitizers in the future. Link: https://git.kernel.org/linus/68df3755e383e6fecf2354a67b08f92f18536594 [1] Link: KSPP#26 [2] Link: KSPP#27 [3] Link: KSPP#344 [4] Cc: Nick Terrell <terrelln@fb.com> Cc: Paul Jones <paul@pauljones.id.au> Cc: Sedat Dilek <sedat.dilek@gmail.com> Cc: Oleksandr Natalenko <oleksandr@natalenko.name> Cc: Xin Gao <gaoxin@cdjrlc.com> Signed-off-by: Kees Cook <keescook@chromium.org>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Much like issue #26 and issue #27, we must mitigate pointer arithmetic wrap-around (overflow/underflow). This should be possible via
-fsanitize=pointer-overflowbut it has similar problems as the other issues, namely-fwrapv-pointer.-fwrapv-pointer(and-fno-strict-overflow)pointer-overflowdoes not appear to function llvm/llvm-project#66451The text was updated successfully, but these errors were encountered: