Chromium sandbox apparently wants to deny statx [1] so it could properly inspect arguments after the sandboxed process later falls back to fstat. Because there's currently not a "fd-only" version of statx, so that the sandbox has no way to ensure the path argument is empty without being able to peek into the sandboxed process's memory. For architectures able to do newfstatat though, glibc falls back to newfstatat after getting -ENOSYS for statx, then the respective SIGSYS handler [2] takes care of inspecting the path argument, transforming allowed newfstatat's into fstat instead which is allowed and has the same type of return value.
But, as LoongArch is the first architecture to not have fstat nor newfstatat, the LoongArch glibc does not attempt falling back at all when it gets -ENOSYS for statx -- and you see the problem there!
Actually, back when the LoongArch port was under review, people were aware of the same problem with sandboxing clone3 [3], so clone was eventually kept. Unfortunately it seemed at that time no one had noticed statx, so besides restoring fstat/newfstatat to LoongArch uapi (and postponing the problem further), it seems inevitable that we would need to tackle seccomp deep argument inspection.
However, this is obviously a decision that shouldn't be taken lightly, so we just restore fstat/newfstatat by defining __ARCH_WANT_NEW_STAT in unistd.h. This is the simplest solution for now, and so we hope the community will tackle the long-standing problem of seccomp deep argument inspection in the future [4][5].
More infomation please reading this thread [6].
[1] https://chromium-review.googlesource.com/c/chromium/src/+/2823150 [2] https://chromium.googlesource.com/chromium/src/sandbox/+/c085b51940bd/linux/... [3] https://lore.kernel.org/linux-arch/20220511211231.GG7074@brightrain.aerifal.... [4] https://lwn.net/Articles/799557/ [5] https://lpc.events/event/4/contributions/560/attachments/397/640/deep-arg-in... [6] https://lore.kernel.org/loongarch/20240226-granit-seilschaft-eccc2433014d@br...
Cc: stable@vger.kernel.org Signed-off-by: Huacai Chen chenhuacai@loongson.cn --- arch/loongarch/include/uapi/asm/unistd.h | 1 + 1 file changed, 1 insertion(+)
diff --git a/arch/loongarch/include/uapi/asm/unistd.h b/arch/loongarch/include/uapi/asm/unistd.h index fcb668984f03..b344b1f91715 100644 --- a/arch/loongarch/include/uapi/asm/unistd.h +++ b/arch/loongarch/include/uapi/asm/unistd.h @@ -1,4 +1,5 @@ /* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */ +#define __ARCH_WANT_NEW_STAT #define __ARCH_WANT_SYS_CLONE #define __ARCH_WANT_SYS_CLONE3
On Sat, May 11, 2024, at 12:01, Huacai Chen wrote:
Chromium sandbox apparently wants to deny statx [1] so it could properly inspect arguments after the sandboxed process later falls back to fstat. Because there's currently not a "fd-only" version of statx, so that the sandbox has no way to ensure the path argument is empty without being able to peek into the sandboxed process's memory. For architectures able to do newfstatat though, glibc falls back to newfstatat after getting -ENOSYS for statx, then the respective SIGSYS handler [2] takes care of inspecting the path argument, transforming allowed newfstatat's into fstat instead which is allowed and has the same type of return value.
But, as LoongArch is the first architecture to not have fstat nor newfstatat, the LoongArch glibc does not attempt falling back at all when it gets -ENOSYS for statx -- and you see the problem there!
My main objection here is that this is inconsistent with 32-bit architectures: we normally have newfstatat() on 64-bit architectures but fstatat64() on 32-bit ones. While loongarch64 is the first 64-bit one that is missing newfstatat(), we have riscv32 already without fstatat64().
Importantly, we can't just add fstatat64() on riscv32 because there is no time64 version for it other than statx(), and I don't want the architectures to diverge more than necessary. I would not mind adding a variant of statx() that works for both riscv32 and loongarch64 though, if it gets added to all architectures.
Arnd
Hi, Arnd,
On Sat, May 11, 2024 at 8:17 PM Arnd Bergmann arnd@arndb.de wrote:
On Sat, May 11, 2024, at 12:01, Huacai Chen wrote:
Chromium sandbox apparently wants to deny statx [1] so it could properly inspect arguments after the sandboxed process later falls back to fstat. Because there's currently not a "fd-only" version of statx, so that the sandbox has no way to ensure the path argument is empty without being able to peek into the sandboxed process's memory. For architectures able to do newfstatat though, glibc falls back to newfstatat after getting -ENOSYS for statx, then the respective SIGSYS handler [2] takes care of inspecting the path argument, transforming allowed newfstatat's into fstat instead which is allowed and has the same type of return value.
But, as LoongArch is the first architecture to not have fstat nor newfstatat, the LoongArch glibc does not attempt falling back at all when it gets -ENOSYS for statx -- and you see the problem there!
My main objection here is that this is inconsistent with 32-bit architectures: we normally have newfstatat() on 64-bit architectures but fstatat64() on 32-bit ones. While loongarch64 is the first 64-bit one that is missing newfstatat(), we have riscv32 already without fstatat64().
Then how to move forward? Xuerui said that he wants to improve seccomp, but a long time has already passed. And I think we should solve this problem before Debian loong64 ports become usable.
Importantly, we can't just add fstatat64() on riscv32 because there is no time64 version for it other than statx(), and I don't want the architectures to diverge more than necessary. I would not mind adding a variant of statx() that works for both riscv32 and loongarch64 though, if it gets added to all architectures.
As far as I know, Ren Guo is trying to implement riscv64 kernel + riscv32 userspace, so I think riscv32 kernel won't be widely used?
Huacai
Arnd
On Sat, May 11, 2024, at 16:28, Huacai Chen wrote:
On Sat, May 11, 2024 at 8:17 PM Arnd Bergmann arnd@arndb.de wrote:
Importantly, we can't just add fstatat64() on riscv32 because there is no time64 version for it other than statx(), and I don't want the architectures to diverge more than necessary. I would not mind adding a variant of statx() that works for both riscv32 and loongarch64 though, if it gets added to all architectures.
As far as I know, Ren Guo is trying to implement riscv64 kernel + riscv32 userspace, so I think riscv32 kernel won't be widely used?
I was talking about the ABI, so it doesn't actually matter what the kernel is: any userspace ABI without CONFIG_COMPAT_32BIT_TIME is equally affected here. On riscv32 this is the only allowed configuration, while on others (arm32 or x86-32 userland) you can turn off COMPAT_32BIT_TIME on both 32-bit kernel and on 64-bit kernels with compat mode.
Arnd
On Sat, May 11, 2024 at 11:39 PM Arnd Bergmann arnd@arndb.de wrote:
On Sat, May 11, 2024, at 16:28, Huacai Chen wrote:
On Sat, May 11, 2024 at 8:17 PM Arnd Bergmann arnd@arndb.de wrote:
Importantly, we can't just add fstatat64() on riscv32 because there is no time64 version for it other than statx(), and I don't want the architectures to diverge more than necessary. I would not mind adding a variant of statx() that works for both riscv32 and loongarch64 though, if it gets added to all architectures.
As far as I know, Ren Guo is trying to implement riscv64 kernel + riscv32 userspace, so I think riscv32 kernel won't be widely used?
I was talking about the ABI, so it doesn't actually matter what the kernel is: any userspace ABI without CONFIG_COMPAT_32BIT_TIME is equally affected here. On riscv32 this is the only allowed configuration, while on others (arm32 or x86-32 userland) you can turn off COMPAT_32BIT_TIME on both 32-bit kernel and on 64-bit kernels with compat mode.
I don't know too much detail, but I think riscv32 can do something similar to arm32 and x86-32, or we can wait for Xuerui to improve seccomp. But there is no much time for loongarch because the Debian loong64 port is coming soon.
Huacai
Arnd
On Sun, May 12, 2024, at 05:11, Huacai Chen wrote:
On Sat, May 11, 2024 at 11:39 PM Arnd Bergmann arnd@arndb.de wrote:
On Sat, May 11, 2024, at 16:28, Huacai Chen wrote:
On Sat, May 11, 2024 at 8:17 PM Arnd Bergmann arnd@arndb.de wrote:
CONFIG_COMPAT_32BIT_TIME is equally affected here. On riscv32 this is the only allowed configuration, while on others (arm32 or x86-32 userland) you can turn off COMPAT_32BIT_TIME on both 32-bit kernel and on 64-bit kernels with compat mode.
I don't know too much detail, but I think riscv32 can do something similar to arm32 and x86-32, or we can wait for Xuerui to improve seccomp. But there is no much time for loongarch because the Debian loong64 port is coming soon.
What I meant is that the other architectures only work by accident if COMPAT_32BIT_TIME is enabled and statx() gets blocked, but then they truncate the timestamps to the tim32 range, which is not acceptable behavior. Actually mips64 is in the same situation because it also only supports 32-bit timestamps in newstatat(), despite being a 64-bit architecture with a 64-bit time_t in all other syscalls.
Arnd
Hi, Arnd,
On Sun, May 12, 2024 at 3:53 PM Arnd Bergmann arnd@arndb.de wrote:
On Sun, May 12, 2024, at 05:11, Huacai Chen wrote:
On Sat, May 11, 2024 at 11:39 PM Arnd Bergmann arnd@arndb.de wrote:
On Sat, May 11, 2024, at 16:28, Huacai Chen wrote:
On Sat, May 11, 2024 at 8:17 PM Arnd Bergmann arnd@arndb.de wrote:
CONFIG_COMPAT_32BIT_TIME is equally affected here. On riscv32 this is the only allowed configuration, while on others (arm32 or x86-32 userland) you can turn off COMPAT_32BIT_TIME on both 32-bit kernel and on 64-bit kernels with compat mode.
I don't know too much detail, but I think riscv32 can do something similar to arm32 and x86-32, or we can wait for Xuerui to improve seccomp. But there is no much time for loongarch because the Debian loong64 port is coming soon.
What I meant is that the other architectures only work by accident if COMPAT_32BIT_TIME is enabled and statx() gets blocked, but then they truncate the timestamps to the tim32 range, which is not acceptable behavior. Actually mips64 is in the same situation because it also only supports 32-bit timestamps in newstatat(), despite being a 64-bit architecture with a 64-bit time_t in all other syscalls.
We can only wait for the seccomp side to be fixed now? Or we can get this patch upstream for LoongArch64 at the moment, and wait for seccomp to fix RISCV32 (and LoongArch32) in future?
Huacai
Arnd
On Sat, 2024-06-15 at 16:52 +0800, Huacai Chen wrote:
Hi, Arnd,
On Sun, May 12, 2024 at 3:53 PM Arnd Bergmann arnd@arndb.de wrote:
On Sun, May 12, 2024, at 05:11, Huacai Chen wrote:
On Sat, May 11, 2024 at 11:39 PM Arnd Bergmann arnd@arndb.de wrote:
On Sat, May 11, 2024, at 16:28, Huacai Chen wrote:
On Sat, May 11, 2024 at 8:17 PM Arnd Bergmann arnd@arndb.de wrote:
CONFIG_COMPAT_32BIT_TIME is equally affected here. On riscv32 this is the only allowed configuration, while on others (arm32 or x86-32 userland) you can turn off COMPAT_32BIT_TIME on both 32-bit kernel and on 64-bit kernels with compat mode.
I don't know too much detail, but I think riscv32 can do something similar to arm32 and x86-32, or we can wait for Xuerui to improve seccomp. But there is no much time for loongarch because the Debian loong64 port is coming soon.
What I meant is that the other architectures only work by accident if COMPAT_32BIT_TIME is enabled and statx() gets blocked, but then they truncate the timestamps to the tim32 range, which is not acceptable behavior. Actually mips64 is in the same situation because it also only supports 32-bit timestamps in newstatat(), despite being a 64-bit architecture with a 64-bit time_t in all other syscalls.
We can only wait for the seccomp side to be fixed now? Or we can get this patch upstream for LoongArch64 at the moment, and wait for seccomp to fix RISCV32 (and LoongArch32) in future?
I'm wondering why not just introduce a new syscall or extend statx with a new flag, as we've discussed many times. They have their own disadvantages but better than this, IMO.
On Sat, Jun 15, 2024 at 4:55 PM Xi Ruoyao xry111@xry111.site wrote:
On Sat, 2024-06-15 at 16:52 +0800, Huacai Chen wrote:
Hi, Arnd,
On Sun, May 12, 2024 at 3:53 PM Arnd Bergmann arnd@arndb.de wrote:
On Sun, May 12, 2024, at 05:11, Huacai Chen wrote:
On Sat, May 11, 2024 at 11:39 PM Arnd Bergmann arnd@arndb.de wrote:
On Sat, May 11, 2024, at 16:28, Huacai Chen wrote:
On Sat, May 11, 2024 at 8:17 PM Arnd Bergmann arnd@arndb.de wrote:
CONFIG_COMPAT_32BIT_TIME is equally affected here. On riscv32 this is the only allowed configuration, while on others (arm32 or x86-32 userland) you can turn off COMPAT_32BIT_TIME on both 32-bit kernel and on 64-bit kernels with compat mode.
I don't know too much detail, but I think riscv32 can do something similar to arm32 and x86-32, or we can wait for Xuerui to improve seccomp. But there is no much time for loongarch because the Debian loong64 port is coming soon.
What I meant is that the other architectures only work by accident if COMPAT_32BIT_TIME is enabled and statx() gets blocked, but then they truncate the timestamps to the tim32 range, which is not acceptable behavior. Actually mips64 is in the same situation because it also only supports 32-bit timestamps in newstatat(), despite being a 64-bit architecture with a 64-bit time_t in all other syscalls.
We can only wait for the seccomp side to be fixed now? Or we can get this patch upstream for LoongArch64 at the moment, and wait for seccomp to fix RISCV32 (and LoongArch32) in future?
I'm wondering why not just introduce a new syscall or extend statx with a new flag, as we've discussed many times. They have their own disadvantages but better than this, IMO.
We should move things forward, in any way. :)
Huacai
-- Xi Ruoyao xry111@xry111.site School of Aerospace Science and Technology, Xidian University
On Sat, Jun 15, 2024, at 11:29, Huacai Chen wrote:
On Sat, Jun 15, 2024 at 4:55 PM Xi Ruoyao xry111@xry111.site wrote:
On Sat, 2024-06-15 at 16:52 +0800, Huacai Chen wrote:
Hi, Arnd,
On Sun, May 12, 2024 at 3:53 PM Arnd Bergmann arnd@arndb.de wrote:
On Sun, May 12, 2024, at 05:11, Huacai Chen wrote:
On Sat, May 11, 2024 at 11:39 PM Arnd Bergmann arnd@arndb.de wrote:
On Sat, May 11, 2024, at 16:28, Huacai Chen wrote: > On Sat, May 11, 2024 at 8:17 PM Arnd Bergmann arnd@arndb.de wrote: CONFIG_COMPAT_32BIT_TIME is equally affected here. On riscv32 this is the only allowed configuration, while on others (arm32 or x86-32 userland) you can turn off COMPAT_32BIT_TIME on both 32-bit kernel and on 64-bit kernels with compat mode.
I don't know too much detail, but I think riscv32 can do something similar to arm32 and x86-32, or we can wait for Xuerui to improve seccomp. But there is no much time for loongarch because the Debian loong64 port is coming soon.
What I meant is that the other architectures only work by accident if COMPAT_32BIT_TIME is enabled and statx() gets blocked, but then they truncate the timestamps to the tim32 range, which is not acceptable behavior. Actually mips64 is in the same situation because it also only supports 32-bit timestamps in newstatat(), despite being a 64-bit architecture with a 64-bit time_t in all other syscalls.
We can only wait for the seccomp side to be fixed now? Or we can get this patch upstream for LoongArch64 at the moment, and wait for seccomp to fix RISCV32 (and LoongArch32) in future?
I'm wondering why not just introduce a new syscall or extend statx with a new flag, as we've discussed many times. They have their own disadvantages but better than this, IMO.
We should move things forward, in any way. :)
Wouldn't it be sufficient to move the AT_EMPTY_PATH hack from vfs_fstatat() to vfs_statx() so we can make them behave the same way?
As far as I can tell, the only difference between the two is that fstatat64() and similar already has added the check for zero-length strings in order to make using vfs_fstatat() fast and safe when called from glibc stat().
Arnd
On Sat, 2024-06-15 at 13:47 +0200, Arnd Bergmann wrote:
/* snip */
We can only wait for the seccomp side to be fixed now? Or we can get this patch upstream for LoongArch64 at the moment, and wait for seccomp to fix RISCV32 (and LoongArch32) in future?
I'm wondering why not just introduce a new syscall or extend statx with a new flag, as we've discussed many times. They have their own disadvantages but better than this, IMO.
We should move things forward, in any way. :)
Wouldn't it be sufficient to move the AT_EMPTY_PATH hack from vfs_fstatat() to vfs_statx() so we can make them behave the same way?
As far as I can tell, the only difference between the two is that fstatat64() and similar already has added the check for zero-length strings in order to make using vfs_fstatat() fast and safe when called from glibc stat().
Do you mean https://git.kernel.org/torvalds/c/9013c51c630a? It (only partially) fix the performance issue but it won't help seccomp. The problem is you cannot check if the string is zero-length with seccomp. Thus seccomp cannot audit fstatat properly as well.
In [Firefox] *all* fstatat (and statx) calls are trapped and *the signal handler* audit this fstatat call. If flags & AT_EMPTY_PATH and path is zero-length, it calls fstat to do the job. But on LoongArch there is no way to "do the job" as the only stat-family call is statx.
[Firefox]:https://searchfox.org/mozilla-central/source/security/sandbox/linux/SandboxF...
On Sat, 2024-06-15 at 20:12 +0800, Xi Ruoyao wrote:
On Sat, 2024-06-15 at 13:47 +0200, Arnd Bergmann wrote:
/* snip */
We can only wait for the seccomp side to be fixed now? Or we can get this patch upstream for LoongArch64 at the moment, and wait for seccomp to fix RISCV32 (and LoongArch32) in future?
I'm wondering why not just introduce a new syscall or extend statx with a new flag, as we've discussed many times. They have their own disadvantages but better than this, IMO.
We should move things forward, in any way. :)
Wouldn't it be sufficient to move the AT_EMPTY_PATH hack from vfs_fstatat() to vfs_statx() so we can make them behave the same way?
As far as I can tell, the only difference between the two is that fstatat64() and similar already has added the check for zero-length strings in order to make using vfs_fstatat() fast and safe when called from glibc stat().
Do you mean https://git.kernel.org/torvalds/c/9013c51c630a?%C2%A0 It (only partially) fix the performance issue but it won't help seccomp. The problem is you cannot check if the string is zero-length with seccomp. Thus seccomp cannot audit fstatat properly as well.
In [Firefox] *all* fstatat (and statx) calls are trapped and *the signal handler* audit this fstatat call. If flags & AT_EMPTY_PATH and path is zero-length, it calls fstat to do the job. But on LoongArch there is no way to "do the job" as the only stat-family call is statx.
Just spent some brain cycles to make a quick hack adding a new statx flag. Patch attached.
On Sat, Jun 15, 2024, at 15:12, Xi Ruoyao wrote:
On Sat, 2024-06-15 at 20:12 +0800, Xi Ruoyao wrote:
Just spent some brain cycles to make a quick hack adding a new statx flag. Patch attached.
Thanks for the prototype. I agree that this is not a good API but that it would address the issue and I am fine with merging something like this if you can convince the VFS maintainers.
Arnd
On Mon, 2024-06-17 at 08:35 +0200, Arnd Bergmann wrote:
On Sat, Jun 15, 2024, at 15:12, Xi Ruoyao wrote:
On Sat, 2024-06-15 at 20:12 +0800, Xi Ruoyao wrote:
x/SandboxFilter.cpp#364
Just spent some brain cycles to make a quick hack adding a new statx flag. Patch attached.
Thanks for the prototype. I agree that this is not a good API
What is particular bad with it? Maybe we can improve before annoying VFS guys :).
but that it would address the issue and I am fine with merging something like this if you can convince the VFS maintainers.
Before that I'd like someone to purpose a better name. I really dislike "AT_FORCE_EMPTY_PATH" but I cannot come up with something better.
On Mon, Jun 17, 2024, at 08:45, Xi Ruoyao wrote:
On Mon, 2024-06-17 at 08:35 +0200, Arnd Bergmann wrote:
On Sat, Jun 15, 2024, at 15:12, Xi Ruoyao wrote:
On Sat, 2024-06-15 at 20:12 +0800, Xi Ruoyao wrote:
x/SandboxFilter.cpp#364
Just spent some brain cycles to make a quick hack adding a new statx flag. Patch attached.
Thanks for the prototype. I agree that this is not a good API
What is particular bad with it? Maybe we can improve before annoying VFS guys :).
I can't come up with anything better either, the problem I see is mainly that the man page has to explain both AT_EMPTY_PATH and AT_FORCE_EMPTY_PATH, which are very similar for compatibility reasons only. We would clearly not design a new interface to have both, but we can't change existing behavior either.
Arnd
Hi, Ruoyao,
On Mon, Jun 17, 2024 at 2:45 PM Xi Ruoyao xry111@xry111.site wrote:
On Mon, 2024-06-17 at 08:35 +0200, Arnd Bergmann wrote:
On Sat, Jun 15, 2024, at 15:12, Xi Ruoyao wrote:
On Sat, 2024-06-15 at 20:12 +0800, Xi Ruoyao wrote:
x/SandboxFilter.cpp#364
Just spent some brain cycles to make a quick hack adding a new statx flag. Patch attached.
Thanks for the prototype. I agree that this is not a good API
What is particular bad with it? Maybe we can improve before annoying VFS guys :).
but that it would address the issue and I am fine with merging something like this if you can convince the VFS maintainers.
Before that I'd like someone to purpose a better name. I really dislike "AT_FORCE_EMPTY_PATH" but I cannot come up with something better.
Any updates? Have you submitted this patch? I hope we can end up at 6.11. :)
Huacai
-- Xi Ruoyao xry111@xry111.site School of Aerospace Science and Technology, Xidian University
On Sat, 2024-06-22 at 15:45 +0800, Huacai Chen wrote:
Hi, Ruoyao,
On Mon, Jun 17, 2024 at 2:45 PM Xi Ruoyao xry111@xry111.site wrote:
On Mon, 2024-06-17 at 08:35 +0200, Arnd Bergmann wrote:
On Sat, Jun 15, 2024, at 15:12, Xi Ruoyao wrote:
On Sat, 2024-06-15 at 20:12 +0800, Xi Ruoyao wrote:
x/SandboxFilter.cpp#364
Just spent some brain cycles to make a quick hack adding a new statx flag. Patch attached.
Thanks for the prototype. I agree that this is not a good API
What is particular bad with it? Maybe we can improve before annoying VFS guys :).
but that it would address the issue and I am fine with merging something like this if you can convince the VFS maintainers.
Before that I'd like someone to purpose a better name. I really dislike "AT_FORCE_EMPTY_PATH" but I cannot come up with something better.
Any updates? Have you submitted this patch? I hope we can end up at 6.11. :)
https://lore.kernel.org/linux-fsdevel/20240622105621.7922-1-xry111@xry111.si...
On 2024/5/11 下午8:17, Arnd Bergmann wrote:
On Sat, May 11, 2024, at 12:01, Huacai Chen wrote:
Chromium sandbox apparently wants to deny statx [1] so it could properly inspect arguments after the sandboxed process later falls back to fstat. Because there's currently not a "fd-only" version of statx, so that the sandbox has no way to ensure the path argument is empty without being able to peek into the sandboxed process's memory. For architectures able to do newfstatat though, glibc falls back to newfstatat after getting -ENOSYS for statx, then the respective SIGSYS handler [2] takes care of inspecting the path argument, transforming allowed newfstatat's into fstat instead which is allowed and has the same type of return value.
But, as LoongArch is the first architecture to not have fstat nor newfstatat, the LoongArch glibc does not attempt falling back at all when it gets -ENOSYS for statx -- and you see the problem there!
My main objection here is that this is inconsistent with 32-bit architectures: we normally have newfstatat() on 64-bit architectures but fstatat64() on 32-bit ones. While loongarch64 is the first 64-bit one that is missing newfstatat(), we have riscv32 already without fstatat64().
Importantly, we can't just add fstatat64() on riscv32 because there is no time64 version for it other than statx(), and I don't want the architectures to diverge more than necessary.
yes, I agree. Normally there is newfstatat() on 64-bit architectures but fstatat64() on 32-bit ones.
I do not understand why fstatat64() can be added for riscv32 still. 32bit timestamp seems works well for the present, it is valid until (0x1UL << 32) / 365 / 24 / 3600 + 1970 == 2106 year. Year 2106 should be enough for 32bit system.
Regards Bibo Mao
I would not mind adding a variant of statx() that works for both riscv32 and loongarch64 though, if it gets added to all architectures.
Arnd
On Wed, May 15, 2024, at 09:30, maobibo wrote:
On 2024/5/11 下午8:17, Arnd Bergmann wrote:
On Sat, May 11, 2024, at 12:01, Huacai Chen wrote:
Importantly, we can't just add fstatat64() on riscv32 because there is no time64 version for it other than statx(), and I don't want the architectures to diverge more than necessary.
yes, I agree. Normally there is newfstatat() on 64-bit architectures but fstatat64() on 32-bit ones.
I do not understand why fstatat64() can be added for riscv32 still. 32bit timestamp seems works well for the present, it is valid until (0x1UL << 32) / 365 / 24 / 3600 + 1970 == 2106 year. Year 2106 should be enough for 32bit system.
There is a very small number of interfaces for which we ended up not using a 64-bit time_t replacement, but those are only for relative times, not epoch based offsets. The main problems here are:
- time_t is defined to be a signed value in posix, and we need to handle file timestamps before 1970 in stat(), so changing this one to be unsigned is not an option.
- A lot of products have already shipped that will have to be supported past 2038 on existing 32-bit hardware. We cannot regress on architectures that have already been fixed to support this.
- file timestamps can also be set into the future, so applications relying on this are broken before 2038.
Arnd
On 2024/5/15 下午10:25, Arnd Bergmann wrote:
On Wed, May 15, 2024, at 09:30, maobibo wrote:
On 2024/5/11 下午8:17, Arnd Bergmann wrote:
On Sat, May 11, 2024, at 12:01, Huacai Chen wrote:
Importantly, we can't just add fstatat64() on riscv32 because there is no time64 version for it other than statx(), and I don't want the architectures to diverge more than necessary.
yes, I agree. Normally there is newfstatat() on 64-bit architectures but fstatat64() on 32-bit ones.
I do not understand why fstatat64() can be added for riscv32 still. 32bit timestamp seems works well for the present, it is valid until (0x1UL << 32) / 365 / 24 / 3600 + 1970 == 2106 year. Year 2106 should be enough for 32bit system.
There is a very small number of interfaces for which we ended up not using a 64-bit time_t replacement, but those are only for relative times, not epoch based offsets. The main problems here are:
time_t is defined to be a signed value in posix, and we need to handle file timestamps before 1970 in stat(), so changing this one to be unsigned is not an option.
A lot of products have already shipped that will have to be supported past 2038 on existing 32-bit hardware. We cannot regress on architectures that have already been fixed to support this.
file timestamps can also be set into the future, so applications relying on this are broken before 2038.
I see. And thanks for detailed explanation.
Regards Bibo Mao
Arnd
linux-stable-mirror@lists.linaro.org
-
Arnd Bergmann -
Huacai Chen -
Huacai Chen -
maobibo -
Xi Ruoyao