答复: 答复: [PATCH] arm64/ptrace: add PTRACE_SYSEMU and PTRACE_SYSEMU_SINGLESTEP support
On Wed, Sep 05, 2018 at 03:45:51AM +0200, Richard Weinberger wrote: > Am Dienstag, 4. September 2018, 04:11:07 CEST schrieb Haibo Xu (Arm > Technology China): > > Hi Richard, > > > > What do you mean by done it in the core? moving macro definition to > > include/uapi/linux/ptrace.h? > >The patch is strictly follow x86's sematic on PTRACE_SYSEMU/SINGLESTEP > >support. > > Well, the feature itself is not really architecture specific. > Just because x86 does it in arch/x86, it does not mean that this is the best > way. > I guess this is also what Will tried to say. > If we can, we should implement PTRACE_SYSEMU in the core ptrace code and not > per architecture. > After re-evaluating the workload to unify the PTRACE_SYSEMU in the core ptrace code on both ARM and x86 architecture, We think we are not able to work on it recently(we only have 1.5 person work on the Gvisor project). > > > > I wonder what Haibo Xu want to do with PTRACE_SYSEMU on arm64. > > > > Are you porting UML or gvisor to arm64? > > > > > > That's a good question. Haibo? > > > > The story is we are working on a container runtime(Google Gvisor) > > support on ARM64 platform, and the Gvisor depend on Linux kernel > > PTRACE_SYSEMU/SINGLESTEP support. > > Gvisor also supports a kvm backend which should be *much* faster than > PTRACE_SYSEMU. > Otherwise gvisor suffers from the same performance drawbacks as UML does. > Pagefaults via SIGSEGV/mmap, syscall gate via ptrace(). > > Did you check, is PTRACE_SYSEMU really the way to go for gvisor? > Last time I checked the KVM backend looked promising but still WIP, though. > Gvisor do support two platforms, ptrace and kvm. To support ptrace platform on ARM64 is our first step. From the long run, KVM would be a better choice, and we will work on it after ptrace platform get worked. Currently, our initial patches for Gvisor ARM64 support are ready for the small "hello-world" container (please refer to the attachment for the test logs), and we are in the process of upstreaming the code. Could you help merge this patch to the Linux kernel for the time being? We can reschedule the resource to unify the overall Ptrace codes in the kernel late. > I also wonder whether PTRACE_SYSEMU is really the only missing bit to support > gvisor on arm64. Did you check how to work around VIPT/VIVT caching issues? > UML (and gvisor in this context) have lots of implicit x86 dependencies. As far as we know, PTRACE_SYSEMU is the only missing bit that needed in Linux kernel to support Gvisor ptrace platform on ARM64. Thanks, Haibo -邮件原件- 发件人: Richard Weinberger 发送时间: 2018年9月5日 3:45 收件人: Haibo Xu (Arm Technology China) 抄送: Will Deacon ; Catalin Marinas ; linux-arm-ker...@lists.infradead.org; linux-kernel@vger.kernel.org; nd ; jd...@addtoit.com 主题: Re: 答复: [PATCH] arm64/ptrace: add PTRACE_SYSEMU and PTRACE_SYSEMU_SINGLESTEP support Am Dienstag, 4. September 2018, 04:11:07 CEST schrieb Haibo Xu (Arm Technology China): > Hi Richard, > > What do you mean by done it in the core? moving macro definition to > include/uapi/linux/ptrace.h? > The patch is strictly follow x86's sematic on PTRACE_SYSEMU/SINGLESTEP > support. Well, the feature itself is not really architecture specific. Just because x86 does it in arch/x86, it does not mean that this is the best way. I guess this is also what Will tried to say. If we can, we should implement PTRACE_SYSEMU in the core ptrace code and not per architecture. > > > I wonder what Haibo Xu want to do with PTRACE_SYSEMU on arm64. > > > Are you porting UML or gvisor to arm64? > > > > That's a good question. Haibo? > > The story is we are working on a container runtime(Google Gvisor) > support on ARM64 platform, and the Gvisor depend on Linux kernel > PTRACE_SYSEMU/SINGLESTEP support. Gvisor also supports a kvm backend which should be *much* faster than PTRACE_SYSEMU. Otherwise gvisor suffers from the same performance drawbacks as UML does. Pagefaults via SIGSEGV/mmap, syscall gate via ptrace(). Did you check, is PTRACE_SYSEMU really the way to go for gvisor? Last time I checked the KVM backend looked promising but still WIP, though. I also wonder whether PTRACE_SYSEMU is really the only missing bit to support gvisor on arm64. Did you check how to work around VIPT/VIVT caching issues? UML (and gvisor in this context) have lots of implicit x86 dependencies. Thanks, //richard IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you. [?201
答复: 答复: [PATCH] arm64/ptrace: add PTRACE_SYSEMU and PTRACE_SYSEMU_SINGLESTEP support
On Wed, Sep 05, 2018 at 03:45:51AM +0200, Richard Weinberger wrote: > Am Dienstag, 4. September 2018, 04:11:07 CEST schrieb Haibo Xu (Arm > Technology China): > > Hi Richard, > > > > What do you mean by done it in the core? moving macro definition to > > include/uapi/linux/ptrace.h? > >The patch is strictly follow x86's sematic on PTRACE_SYSEMU/SINGLESTEP > >support. > > Well, the feature itself is not really architecture specific. > Just because x86 does it in arch/x86, it does not mean that this is the best > way. > I guess this is also what Will tried to say. > If we can, we should implement PTRACE_SYSEMU in the core ptrace code and not > per architecture. > After re-evaluating the workload to unify the PTRACE_SYSEMU in the core ptrace code on both ARM and x86 architecture, We think we are not able to work on it recently(we only have 1.5 person work on the Gvisor project). > > > > I wonder what Haibo Xu want to do with PTRACE_SYSEMU on arm64. > > > > Are you porting UML or gvisor to arm64? > > > > > > That's a good question. Haibo? > > > > The story is we are working on a container runtime(Google Gvisor) > > support on ARM64 platform, and the Gvisor depend on Linux kernel > > PTRACE_SYSEMU/SINGLESTEP support. > > Gvisor also supports a kvm backend which should be *much* faster than > PTRACE_SYSEMU. > Otherwise gvisor suffers from the same performance drawbacks as UML does. > Pagefaults via SIGSEGV/mmap, syscall gate via ptrace(). > > Did you check, is PTRACE_SYSEMU really the way to go for gvisor? > Last time I checked the KVM backend looked promising but still WIP, though. > Gvisor do support two platforms, ptrace and kvm. To support ptrace platform on ARM64 is our first step. From the long run, KVM would be a better choice, and we will work on it after ptrace platform get worked. Currently, our initial patches for Gvisor ARM64 support are ready for the small "hello-world" container (please refer to the attachment for the test logs), and we are in the process of upstreaming the code. Could you help merge this patch to the Linux kernel for the time being? We can reschedule the resource to unify the overall Ptrace codes in the kernel late. > I also wonder whether PTRACE_SYSEMU is really the only missing bit to support > gvisor on arm64. Did you check how to work around VIPT/VIVT caching issues? > UML (and gvisor in this context) have lots of implicit x86 dependencies. As far as we know, PTRACE_SYSEMU is the only missing bit that needed in Linux kernel to support Gvisor ptrace platform on ARM64. Thanks, Haibo -邮件原件- 发件人: Richard Weinberger 发送时间: 2018年9月5日 3:45 收件人: Haibo Xu (Arm Technology China) 抄送: Will Deacon ; Catalin Marinas ; linux-arm-ker...@lists.infradead.org; linux-kernel@vger.kernel.org; nd ; jd...@addtoit.com 主题: Re: 答复: [PATCH] arm64/ptrace: add PTRACE_SYSEMU and PTRACE_SYSEMU_SINGLESTEP support Am Dienstag, 4. September 2018, 04:11:07 CEST schrieb Haibo Xu (Arm Technology China): > Hi Richard, > > What do you mean by done it in the core? moving macro definition to > include/uapi/linux/ptrace.h? > The patch is strictly follow x86's sematic on PTRACE_SYSEMU/SINGLESTEP > support. Well, the feature itself is not really architecture specific. Just because x86 does it in arch/x86, it does not mean that this is the best way. I guess this is also what Will tried to say. If we can, we should implement PTRACE_SYSEMU in the core ptrace code and not per architecture. > > > I wonder what Haibo Xu want to do with PTRACE_SYSEMU on arm64. > > > Are you porting UML or gvisor to arm64? > > > > That's a good question. Haibo? > > The story is we are working on a container runtime(Google Gvisor) > support on ARM64 platform, and the Gvisor depend on Linux kernel > PTRACE_SYSEMU/SINGLESTEP support. Gvisor also supports a kvm backend which should be *much* faster than PTRACE_SYSEMU. Otherwise gvisor suffers from the same performance drawbacks as UML does. Pagefaults via SIGSEGV/mmap, syscall gate via ptrace(). Did you check, is PTRACE_SYSEMU really the way to go for gvisor? Last time I checked the KVM backend looked promising but still WIP, though. I also wonder whether PTRACE_SYSEMU is really the only missing bit to support gvisor on arm64. Did you check how to work around VIPT/VIVT caching issues? UML (and gvisor in this context) have lots of implicit x86 dependencies. Thanks, //richard IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you. [?201
答复: 答复: [PATCH] arm64/ptrace: add PTRACE_SYSEMU and PTRACE_SYSEMU_SINGLESTEP support
On Wed, Sep 05, 2018 at 03:45:51AM +0200, Richard Weinberger wrote: > Am Dienstag, 4. September 2018, 04:11:07 CEST schrieb Haibo Xu (Arm > Technology China): > > Hi Richard, > > > > What do you mean by done it in the core? moving macro definition to > > include/uapi/linux/ptrace.h? > >The patch is strictly follow x86's sematic on PTRACE_SYSEMU/SINGLESTEP > >support. > > Well, the feature itself is not really architecture specific. > Just because x86 does it in arch/x86, it does not mean that this is the best > way. > I guess this is also what Will tried to say. > If we can, we should implement PTRACE_SYSEMU in the core ptrace code and not > per architecture. > Yes, the feature is common on x86/ARM64, and there are many duplicated codes on both architecture specific ptrace codes. But to unify these codes may take more time, we need to re-evaluate the workload. > > > > I wonder what Haibo Xu want to do with PTRACE_SYSEMU on arm64. > > > > Are you porting UML or gvisor to arm64? > > > > > > That's a good question. Haibo? > > > > The story is we are working on a container runtime(Google Gvisor) > > support on ARM64 platform, and the Gvisor depend on Linux kernel > > PTRACE_SYSEMU/SINGLESTEP support. > > Gvisor also supports a kvm backend which should be *much* faster than > PTRACE_SYSEMU. > Otherwise gvisor suffers from the same performance drawbacks as UML does. > Pagefaults via SIGSEGV/mmap, syscall gate via ptrace(). > > Did you check, is PTRACE_SYSEMU really the way to go for gvisor? > Last time I checked the KVM backend looked promising but still WIP, though. > Gvisor do support two platforms, ptrace and kvm. To support ptrace platform on ARM64 is our first step. From the long run, KVM would be a better choice, and we will work on it after ptrace platform get worked. > I also wonder whether PTRACE_SYSEMU is really the only missing bit to support > gvisor on arm64. Did you check how to work around VIPT/VIVT caching issues? > UML (and gvisor in this context) have lots of implicit x86 dependencies. As far as we know, PTRACE_SYSEMU is the only missing bit that needed in Linux kernel to support Gvisor on ARM64. For the VIPT/VIVT caching issues, we haven't look into it. Could you help share more info about the issue? Thanks, Haibo -邮件原件- 发件人: Richard Weinberger 发送时间: 2018年9月5日 3:45 收件人: Haibo Xu (Arm Technology China) 抄送: Will Deacon ; Catalin Marinas ; linux-arm-ker...@lists.infradead.org; linux-kernel@vger.kernel.org; nd ; jd...@addtoit.com 主题: Re: 答复: [PATCH] arm64/ptrace: add PTRACE_SYSEMU and PTRACE_SYSEMU_SINGLESTEP support Am Dienstag, 4. September 2018, 04:11:07 CEST schrieb Haibo Xu (Arm Technology China): > Hi Richard, > > What do you mean by done it in the core? moving macro definition to > include/uapi/linux/ptrace.h? > The patch is strictly follow x86's sematic on PTRACE_SYSEMU/SINGLESTEP > support. Well, the feature itself is not really architecture specific. Just because x86 does it in arch/x86, it does not mean that this is the best way. I guess this is also what Will tried to say. If we can, we should implement PTRACE_SYSEMU in the core ptrace code and not per architecture. > > > I wonder what Haibo Xu want to do with PTRACE_SYSEMU on arm64. > > > Are you porting UML or gvisor to arm64? > > > > That's a good question. Haibo? > > The story is we are working on a container runtime(Google Gvisor) > support on ARM64 platform, and the Gvisor depend on Linux kernel > PTRACE_SYSEMU/SINGLESTEP support. Gvisor also supports a kvm backend which should be *much* faster than PTRACE_SYSEMU. Otherwise gvisor suffers from the same performance drawbacks as UML does. Pagefaults via SIGSEGV/mmap, syscall gate via ptrace(). Did you check, is PTRACE_SYSEMU really the way to go for gvisor? Last time I checked the KVM backend looked promising but still WIP, though. I also wonder whether PTRACE_SYSEMU is really the only missing bit to support gvisor on arm64. Did you check how to work around VIPT/VIVT caching issues? UML (and gvisor in this context) have lots of implicit x86 dependencies. Thanks, //richard IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
答复: 答复: [PATCH] arm64/ptrace: add PTRACE_SYSEMU and PTRACE_SYSEMU_SINGLESTEP support
On Wed, Sep 05, 2018 at 03:45:51AM +0200, Richard Weinberger wrote: > Am Dienstag, 4. September 2018, 04:11:07 CEST schrieb Haibo Xu (Arm > Technology China): > > Hi Richard, > > > > What do you mean by done it in the core? moving macro definition to > > include/uapi/linux/ptrace.h? > >The patch is strictly follow x86's sematic on PTRACE_SYSEMU/SINGLESTEP > >support. > > Well, the feature itself is not really architecture specific. > Just because x86 does it in arch/x86, it does not mean that this is the best > way. > I guess this is also what Will tried to say. > If we can, we should implement PTRACE_SYSEMU in the core ptrace code and not > per architecture. > Yes, the feature is common on x86/ARM64, and there are many duplicated codes on both architecture specific ptrace codes. But to unify these codes may take more time, we need to re-evaluate the workload. > > > > I wonder what Haibo Xu want to do with PTRACE_SYSEMU on arm64. > > > > Are you porting UML or gvisor to arm64? > > > > > > That's a good question. Haibo? > > > > The story is we are working on a container runtime(Google Gvisor) > > support on ARM64 platform, and the Gvisor depend on Linux kernel > > PTRACE_SYSEMU/SINGLESTEP support. > > Gvisor also supports a kvm backend which should be *much* faster than > PTRACE_SYSEMU. > Otherwise gvisor suffers from the same performance drawbacks as UML does. > Pagefaults via SIGSEGV/mmap, syscall gate via ptrace(). > > Did you check, is PTRACE_SYSEMU really the way to go for gvisor? > Last time I checked the KVM backend looked promising but still WIP, though. > Gvisor do support two platforms, ptrace and kvm. To support ptrace platform on ARM64 is our first step. From the long run, KVM would be a better choice, and we will work on it after ptrace platform get worked. > I also wonder whether PTRACE_SYSEMU is really the only missing bit to support > gvisor on arm64. Did you check how to work around VIPT/VIVT caching issues? > UML (and gvisor in this context) have lots of implicit x86 dependencies. As far as we know, PTRACE_SYSEMU is the only missing bit that needed in Linux kernel to support Gvisor on ARM64. For the VIPT/VIVT caching issues, we haven't look into it. Could you help share more info about the issue? Thanks, Haibo -邮件原件- 发件人: Richard Weinberger 发送时间: 2018年9月5日 3:45 收件人: Haibo Xu (Arm Technology China) 抄送: Will Deacon ; Catalin Marinas ; linux-arm-ker...@lists.infradead.org; linux-kernel@vger.kernel.org; nd ; jd...@addtoit.com 主题: Re: 答复: [PATCH] arm64/ptrace: add PTRACE_SYSEMU and PTRACE_SYSEMU_SINGLESTEP support Am Dienstag, 4. September 2018, 04:11:07 CEST schrieb Haibo Xu (Arm Technology China): > Hi Richard, > > What do you mean by done it in the core? moving macro definition to > include/uapi/linux/ptrace.h? > The patch is strictly follow x86's sematic on PTRACE_SYSEMU/SINGLESTEP > support. Well, the feature itself is not really architecture specific. Just because x86 does it in arch/x86, it does not mean that this is the best way. I guess this is also what Will tried to say. If we can, we should implement PTRACE_SYSEMU in the core ptrace code and not per architecture. > > > I wonder what Haibo Xu want to do with PTRACE_SYSEMU on arm64. > > > Are you porting UML or gvisor to arm64? > > > > That's a good question. Haibo? > > The story is we are working on a container runtime(Google Gvisor) > support on ARM64 platform, and the Gvisor depend on Linux kernel > PTRACE_SYSEMU/SINGLESTEP support. Gvisor also supports a kvm backend which should be *much* faster than PTRACE_SYSEMU. Otherwise gvisor suffers from the same performance drawbacks as UML does. Pagefaults via SIGSEGV/mmap, syscall gate via ptrace(). Did you check, is PTRACE_SYSEMU really the way to go for gvisor? Last time I checked the KVM backend looked promising but still WIP, though. I also wonder whether PTRACE_SYSEMU is really the only missing bit to support gvisor on arm64. Did you check how to work around VIPT/VIVT caching issues? UML (and gvisor in this context) have lots of implicit x86 dependencies. Thanks, //richard IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
Re: 答复: [PATCH] arm64/ptrace: add PTRACE_SYSEMU and PTRACE_SYSEMU_SINGLESTEP support
Am Dienstag, 4. September 2018, 04:11:07 CEST schrieb Haibo Xu (Arm Technology China): > Hi Richard, > > What do you mean by done it in the core? moving macro definition to > include/uapi/linux/ptrace.h? > The patch is strictly follow x86's sematic on PTRACE_SYSEMU/SINGLESTEP > support. Well, the feature itself is not really architecture specific. Just because x86 does it in arch/x86, it does not mean that this is the best way. I guess this is also what Will tried to say. If we can, we should implement PTRACE_SYSEMU in the core ptrace code and not per architecture. > > > I wonder what Haibo Xu want to do with PTRACE_SYSEMU on arm64. > > > Are you porting UML or gvisor to arm64? > > > > That's a good question. Haibo? > > The story is we are working on a container runtime(Google Gvisor) support on > ARM64 platform, > and the Gvisor depend on Linux kernel PTRACE_SYSEMU/SINGLESTEP support. Gvisor also supports a kvm backend which should be *much* faster than PTRACE_SYSEMU. Otherwise gvisor suffers from the same performance drawbacks as UML does. Pagefaults via SIGSEGV/mmap, syscall gate via ptrace(). Did you check, is PTRACE_SYSEMU really the way to go for gvisor? Last time I checked the KVM backend looked promising but still WIP, though. I also wonder whether PTRACE_SYSEMU is really the only missing bit to support gvisor on arm64. Did you check how to work around VIPT/VIVT caching issues? UML (and gvisor in this context) have lots of implicit x86 dependencies. Thanks, //richard
Re: 答复: [PATCH] arm64/ptrace: add PTRACE_SYSEMU and PTRACE_SYSEMU_SINGLESTEP support
Am Dienstag, 4. September 2018, 04:11:07 CEST schrieb Haibo Xu (Arm Technology China): > Hi Richard, > > What do you mean by done it in the core? moving macro definition to > include/uapi/linux/ptrace.h? > The patch is strictly follow x86's sematic on PTRACE_SYSEMU/SINGLESTEP > support. Well, the feature itself is not really architecture specific. Just because x86 does it in arch/x86, it does not mean that this is the best way. I guess this is also what Will tried to say. If we can, we should implement PTRACE_SYSEMU in the core ptrace code and not per architecture. > > > I wonder what Haibo Xu want to do with PTRACE_SYSEMU on arm64. > > > Are you porting UML or gvisor to arm64? > > > > That's a good question. Haibo? > > The story is we are working on a container runtime(Google Gvisor) support on > ARM64 platform, > and the Gvisor depend on Linux kernel PTRACE_SYSEMU/SINGLESTEP support. Gvisor also supports a kvm backend which should be *much* faster than PTRACE_SYSEMU. Otherwise gvisor suffers from the same performance drawbacks as UML does. Pagefaults via SIGSEGV/mmap, syscall gate via ptrace(). Did you check, is PTRACE_SYSEMU really the way to go for gvisor? Last time I checked the KVM backend looked promising but still WIP, though. I also wonder whether PTRACE_SYSEMU is really the only missing bit to support gvisor on arm64. Did you check how to work around VIPT/VIVT caching issues? UML (and gvisor in this context) have lots of implicit x86 dependencies. Thanks, //richard
答复: [PATCH] arm64/ptrace: add PTRACE_SYSEMU and PTRACE_SYSEMU_SINGLESTEP support
Hi Will, > On Mon, Sep 03, 2018 at 06:40:51PM +0200, Richard Weinberger wrote: > > Am Montag, 3. September 2018, 18:31:03 CEST schrieb Will Deacon: > > > On Mon, Sep 03, 2018 at 02:23:17PM +0800, Haibo.Xu wrote: > > > > Add PTRACE_SYSEMU and PTRACE_SYSEMU_SINGLESTEP support on ARM64. > > > > This copies the x86 semantics for invoking ptrace hooks, and have > > > > been verified on ARM64 machine. > > > > > > > > Signed-off-by: Haibo.Xu > > > > Signed-off-by: Bin.Lu > > > > --- > > > > arch/arm64/include/asm/thread_info.h | 5 - > > > > arch/arm64/include/uapi/asm/ptrace.h | 2 ++ > > > > arch/arm64/kernel/ptrace.c | 17 + > > > > 3 files changed, 23 insertions(+), 1 deletion(-) > > > > > > What is PTRACE_SYSEMU and what is its semantics? Why isn't it done > > > in the core ptrace code? > > > > It is an optimization added for UserModeLinux many years ago. > > PTRACE_SYSEMU basically allows you to handle system calls in user > > space without the kernel seeing them. > > Before that UML had to render every system call into a no-op, e.h. getpid(). > > This was complicated and slow. > > Ok, thanks for the insight! > Thanks for Richard's comments! > > The ptrace() manpage has a section on PTRACE_SYSEMU, more > > documentation on the semantics is not available. > > :( We already have tracehook, seccomp, tracepoint and audit fighting with > each other on syscall entry, so I'm really not keen to add another player to > the game unless we really have to. > > Has anybody tried implementing this using tracehook? > Currently, it seems nobody has tried it. > > And yes, I think this should also done in the core. Like many other > > ptrace() areas this needs a cleanup. ;-) > > Hi Richard, What do you mean by done it in the core? moving macro definition to include/uapi/linux/ptrace.h? The patch is strictly follow x86's sematic on PTRACE_SYSEMU/SINGLESTEP support. > > I wonder what Haibo Xu want to do with PTRACE_SYSEMU on arm64. > > Are you porting UML or gvisor to arm64? > > That's a good question. Haibo? The story is we are working on a container runtime(Google Gvisor) support on ARM64 platform, and the Gvisor depend on Linux kernel PTRACE_SYSEMU/SINGLESTEP support. Haibo -邮件原件- 发件人: Will Deacon 发送时间: 2018年9月4日 0:57 收件人: Richard Weinberger 抄送: Haibo Xu (Arm Technology China) ; Catalin Marinas ; linux-arm-ker...@lists.infradead.org; linux-kernel@vger.kernel.org; nd ; jd...@addtoit.com 主题: Re: [PATCH] arm64/ptrace: add PTRACE_SYSEMU and PTRACE_SYSEMU_SINGLESTEP support Hi Richard, On Mon, Sep 03, 2018 at 06:40:51PM +0200, Richard Weinberger wrote: > Am Montag, 3. September 2018, 18:31:03 CEST schrieb Will Deacon: > > On Mon, Sep 03, 2018 at 02:23:17PM +0800, Haibo.Xu wrote: > > > Add PTRACE_SYSEMU and PTRACE_SYSEMU_SINGLESTEP support on ARM64. > > > This copies the x86 semantics for invoking ptrace hooks, and have > > > been verified on ARM64 machine. > > > > > > Signed-off-by: Haibo.Xu > > > Signed-off-by: Bin.Lu > > > --- > > > arch/arm64/include/asm/thread_info.h | 5 - > > > arch/arm64/include/uapi/asm/ptrace.h | 2 ++ > > > arch/arm64/kernel/ptrace.c | 17 + > > > 3 files changed, 23 insertions(+), 1 deletion(-) > > > > What is PTRACE_SYSEMU and what is its semantics? Why isn't it done > > in the core ptrace code? > > It is an optimization added for UserModeLinux many years ago. > PTRACE_SYSEMU basically allows you to handle system calls in user > space without the kernel seeing them. > Before that UML had to render every system call into a no-op, e.h. getpid(). > This was complicated and slow. Ok, thanks for the insight! > The ptrace() manpage has a section on PTRACE_SYSEMU, more > documentation on the semantics is not available. :( We already have tracehook, seccomp, tracepoint and audit fighting with each other on syscall entry, so I'm really not keen to add another player to the game unless we really have to. Has anybody tried implementing this using tracehook? > And yes, I think this should also done in the core. Like many other > ptrace() areas this needs a cleanup. ;-) > > I wonder what Haibo Xu want to do with PTRACE_SYSEMU on arm64. > Are you porting UML or gvisor to arm64? That's a good question. Haibo? Will IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
答复: [PATCH] arm64/ptrace: add PTRACE_SYSEMU and PTRACE_SYSEMU_SINGLESTEP support
Hi Will, > On Mon, Sep 03, 2018 at 06:40:51PM +0200, Richard Weinberger wrote: > > Am Montag, 3. September 2018, 18:31:03 CEST schrieb Will Deacon: > > > On Mon, Sep 03, 2018 at 02:23:17PM +0800, Haibo.Xu wrote: > > > > Add PTRACE_SYSEMU and PTRACE_SYSEMU_SINGLESTEP support on ARM64. > > > > This copies the x86 semantics for invoking ptrace hooks, and have > > > > been verified on ARM64 machine. > > > > > > > > Signed-off-by: Haibo.Xu > > > > Signed-off-by: Bin.Lu > > > > --- > > > > arch/arm64/include/asm/thread_info.h | 5 - > > > > arch/arm64/include/uapi/asm/ptrace.h | 2 ++ > > > > arch/arm64/kernel/ptrace.c | 17 + > > > > 3 files changed, 23 insertions(+), 1 deletion(-) > > > > > > What is PTRACE_SYSEMU and what is its semantics? Why isn't it done > > > in the core ptrace code? > > > > It is an optimization added for UserModeLinux many years ago. > > PTRACE_SYSEMU basically allows you to handle system calls in user > > space without the kernel seeing them. > > Before that UML had to render every system call into a no-op, e.h. getpid(). > > This was complicated and slow. > > Ok, thanks for the insight! > Thanks for Richard's comments! > > The ptrace() manpage has a section on PTRACE_SYSEMU, more > > documentation on the semantics is not available. > > :( We already have tracehook, seccomp, tracepoint and audit fighting with > each other on syscall entry, so I'm really not keen to add another player to > the game unless we really have to. > > Has anybody tried implementing this using tracehook? > Currently, it seems nobody has tried it. > > And yes, I think this should also done in the core. Like many other > > ptrace() areas this needs a cleanup. ;-) > > Hi Richard, What do you mean by done it in the core? moving macro definition to include/uapi/linux/ptrace.h? The patch is strictly follow x86's sematic on PTRACE_SYSEMU/SINGLESTEP support. > > I wonder what Haibo Xu want to do with PTRACE_SYSEMU on arm64. > > Are you porting UML or gvisor to arm64? > > That's a good question. Haibo? The story is we are working on a container runtime(Google Gvisor) support on ARM64 platform, and the Gvisor depend on Linux kernel PTRACE_SYSEMU/SINGLESTEP support. Haibo -邮件原件- 发件人: Will Deacon 发送时间: 2018年9月4日 0:57 收件人: Richard Weinberger 抄送: Haibo Xu (Arm Technology China) ; Catalin Marinas ; linux-arm-ker...@lists.infradead.org; linux-kernel@vger.kernel.org; nd ; jd...@addtoit.com 主题: Re: [PATCH] arm64/ptrace: add PTRACE_SYSEMU and PTRACE_SYSEMU_SINGLESTEP support Hi Richard, On Mon, Sep 03, 2018 at 06:40:51PM +0200, Richard Weinberger wrote: > Am Montag, 3. September 2018, 18:31:03 CEST schrieb Will Deacon: > > On Mon, Sep 03, 2018 at 02:23:17PM +0800, Haibo.Xu wrote: > > > Add PTRACE_SYSEMU and PTRACE_SYSEMU_SINGLESTEP support on ARM64. > > > This copies the x86 semantics for invoking ptrace hooks, and have > > > been verified on ARM64 machine. > > > > > > Signed-off-by: Haibo.Xu > > > Signed-off-by: Bin.Lu > > > --- > > > arch/arm64/include/asm/thread_info.h | 5 - > > > arch/arm64/include/uapi/asm/ptrace.h | 2 ++ > > > arch/arm64/kernel/ptrace.c | 17 + > > > 3 files changed, 23 insertions(+), 1 deletion(-) > > > > What is PTRACE_SYSEMU and what is its semantics? Why isn't it done > > in the core ptrace code? > > It is an optimization added for UserModeLinux many years ago. > PTRACE_SYSEMU basically allows you to handle system calls in user > space without the kernel seeing them. > Before that UML had to render every system call into a no-op, e.h. getpid(). > This was complicated and slow. Ok, thanks for the insight! > The ptrace() manpage has a section on PTRACE_SYSEMU, more > documentation on the semantics is not available. :( We already have tracehook, seccomp, tracepoint and audit fighting with each other on syscall entry, so I'm really not keen to add another player to the game unless we really have to. Has anybody tried implementing this using tracehook? > And yes, I think this should also done in the core. Like many other > ptrace() areas this needs a cleanup. ;-) > > I wonder what Haibo Xu want to do with PTRACE_SYSEMU on arm64. > Are you porting UML or gvisor to arm64? That's a good question. Haibo? Will IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
