Re: [PATCH v3 00/31] AArch64 Linux kernel port
On Sat, Sep 08, 2012 at 12:25:37AM +0100, Olof Johansson wrote: > On Fri, Sep 07, 2012 at 05:26:35PM +0100, Catalin Marinas wrote: > > This is the 3rd version of patches implementing Linux kernel support for > > the 64-bit ARM architecture. Most of the review comments for the > > previous release have been implemented. There are a few left (like > > generic timer dts specification) but the main goal was to freeze the > > user ABI. > > > > The Linux kernel patches are available on this tree: > > > > git://git.kernel.org/pub/scm/linux/kernel/git/cmarinas/linux-aarch64.git > > upstream > > > > Main changes since the previous version: > > - ptrace interface now using regsets for TLS and hardware debug > > registers. > > - sys_personality fixes. PER_LINUX is the default for all tasks (32 or > > 64-bit) and there is no compat_sys_personality defined. > > - Kconfig clean-up (unused symbols, AARCH32_EMULATION removed in favour > > of COMPAT). > > - Clean-up of the __cpuc_* function prefixes. > > - CPU information table implemented in C rather than assembly. > > - __bad_* function calls replaced with BUILD_BUG. > > - Inline get_user/put_user (code cleaner, slightly bigger kernel Image). > > - DMA API implementation simplified, defaulting to swiotlb. It will be > > extended as hardware becomes available. > > - Arch-specific code in the generic timer patch moved to arch/arm64/. > > - Comment fixes and clarifications. > > I see that a good amount of the comments I had on v2 have neither been replied > to nor addressed. Sorry, as I said above, my main focus was get the user ABI sorted out as a code is public and there are people that started using it. There are still a few emails marked as "unread" in my inbox that need to be addressed but I did not consider them critical for version 3 of the series. They are not forgotten, however, I'm working back through those emails and either reply or implement the suggestions. > Acked-by: Olof Johansson Thanks. -- Catalin -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH v3 00/31] AArch64 Linux kernel port
On Monday 10 September 2012, Jon Masters wrote: > On 09/07/2012 12:26 PM, Catalin Marinas wrote: > > This is the 3rd version of patches implementing Linux kernel support for > > the 64-bit ARM architecture. > > > > ...but the main goal was to freeze the user ABI. > > Is the consensus now that we're good from an ABI viewpoint? > >From my point of view, no more changes are needed. I would not give any guarantees until we have a first kernel release including this code, but I think we can agree that we would consult you about ABI changes if they seem necessary before then. If you start (re)building the Fedora packages based on these kernel headers, the changes are pretty good that this will keep working. Arnd -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH v3 00/31] AArch64 Linux kernel port
On Monday 10 September 2012, Jon Masters wrote: On 09/07/2012 12:26 PM, Catalin Marinas wrote: This is the 3rd version of patches implementing Linux kernel support for the 64-bit ARM architecture. snip ...but the main goal was to freeze the user ABI. Is the consensus now that we're good from an ABI viewpoint? From my point of view, no more changes are needed. I would not give any guarantees until we have a first kernel release including this code, but I think we can agree that we would consult you about ABI changes if they seem necessary before then. If you start (re)building the Fedora packages based on these kernel headers, the changes are pretty good that this will keep working. Arnd -- To unsubscribe from this list: send the line unsubscribe linux-kernel in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH v3 00/31] AArch64 Linux kernel port
On Sat, Sep 08, 2012 at 12:25:37AM +0100, Olof Johansson wrote: On Fri, Sep 07, 2012 at 05:26:35PM +0100, Catalin Marinas wrote: This is the 3rd version of patches implementing Linux kernel support for the 64-bit ARM architecture. Most of the review comments for the previous release have been implemented. There are a few left (like generic timer dts specification) but the main goal was to freeze the user ABI. The Linux kernel patches are available on this tree: git://git.kernel.org/pub/scm/linux/kernel/git/cmarinas/linux-aarch64.git upstream Main changes since the previous version: - ptrace interface now using regsets for TLS and hardware debug registers. - sys_personality fixes. PER_LINUX is the default for all tasks (32 or 64-bit) and there is no compat_sys_personality defined. - Kconfig clean-up (unused symbols, AARCH32_EMULATION removed in favour of COMPAT). - Clean-up of the __cpuc_* function prefixes. - CPU information table implemented in C rather than assembly. - __bad_* function calls replaced with BUILD_BUG. - Inline get_user/put_user (code cleaner, slightly bigger kernel Image). - DMA API implementation simplified, defaulting to swiotlb. It will be extended as hardware becomes available. - Arch-specific code in the generic timer patch moved to arch/arm64/. - Comment fixes and clarifications. I see that a good amount of the comments I had on v2 have neither been replied to nor addressed. Sorry, as I said above, my main focus was get the user ABI sorted out as a code is public and there are people that started using it. There are still a few emails marked as unread in my inbox that need to be addressed but I did not consider them critical for version 3 of the series. They are not forgotten, however, I'm working back through those emails and either reply or implement the suggestions. Acked-by: Olof Johansson o...@lixom.net Thanks. -- Catalin -- To unsubscribe from this list: send the line unsubscribe linux-kernel in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH v3 00/31] AArch64 Linux kernel port
On 09/07/2012 12:26 PM, Catalin Marinas wrote: This is the 3rd version of patches implementing Linux kernel support for the 64-bit ARM architecture. ...but the main goal was to freeze the user ABI. Is the consensus now that we're good from an ABI viewpoint? Jon. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH v3 00/31] AArch64 Linux kernel port
On Fri, 7 Sep 2012, Catalin Marinas wrote: > This is the 3rd version of patches implementing Linux kernel support for > the 64-bit ARM architecture. Most of the review comments for the > previous release have been implemented. There are a few left (like > generic timer dts specification) but the main goal was to freeze the > user ABI. > > The Linux kernel patches are available on this tree: > > git://git.kernel.org/pub/scm/linux/kernel/git/cmarinas/linux-aarch64.git > upstream > > Main changes since the previous version: > - ptrace interface now using regsets for TLS and hardware debug > registers. > - sys_personality fixes. PER_LINUX is the default for all tasks (32 or > 64-bit) and there is no compat_sys_personality defined. > - Kconfig clean-up (unused symbols, AARCH32_EMULATION removed in favour > of COMPAT). > - Clean-up of the __cpuc_* function prefixes. > - CPU information table implemented in C rather than assembly. > - __bad_* function calls replaced with BUILD_BUG. > - Inline get_user/put_user (code cleaner, slightly bigger kernel Image). > - DMA API implementation simplified, defaulting to swiotlb. It will be > extended as hardware becomes available. > - Arch-specific code in the generic timer patch moved to arch/arm64/. > - Comment fixes and clarifications. For those patches in this series I didn't comment on already: Acked-by: Nicolas Pitre Small nit: please s/n...@cam.org/n...@fluxnic.net/. Nicolas -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH v3 00/31] AArch64 Linux kernel port
On Fri, 7 Sep 2012, Catalin Marinas wrote: This is the 3rd version of patches implementing Linux kernel support for the 64-bit ARM architecture. Most of the review comments for the previous release have been implemented. There are a few left (like generic timer dts specification) but the main goal was to freeze the user ABI. The Linux kernel patches are available on this tree: git://git.kernel.org/pub/scm/linux/kernel/git/cmarinas/linux-aarch64.git upstream Main changes since the previous version: - ptrace interface now using regsets for TLS and hardware debug registers. - sys_personality fixes. PER_LINUX is the default for all tasks (32 or 64-bit) and there is no compat_sys_personality defined. - Kconfig clean-up (unused symbols, AARCH32_EMULATION removed in favour of COMPAT). - Clean-up of the __cpuc_* function prefixes. - CPU information table implemented in C rather than assembly. - __bad_* function calls replaced with BUILD_BUG. - Inline get_user/put_user (code cleaner, slightly bigger kernel Image). - DMA API implementation simplified, defaulting to swiotlb. It will be extended as hardware becomes available. - Arch-specific code in the generic timer patch moved to arch/arm64/. - Comment fixes and clarifications. For those patches in this series I didn't comment on already: Acked-by: Nicolas Pitre n...@linaro.org Small nit: please s/n...@cam.org/n...@fluxnic.net/. Nicolas -- To unsubscribe from this list: send the line unsubscribe linux-kernel in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH v3 00/31] AArch64 Linux kernel port
On 09/07/2012 12:26 PM, Catalin Marinas wrote: This is the 3rd version of patches implementing Linux kernel support for the 64-bit ARM architecture. snip ...but the main goal was to freeze the user ABI. Is the consensus now that we're good from an ABI viewpoint? Jon. -- To unsubscribe from this list: send the line unsubscribe linux-kernel in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH v3 00/31] AArch64 Linux kernel port
On Sat, Sep 8, 2012 at 7:29 PM, Nicolas Pitre wrote: > On Sat, 8 Sep 2012, Santosh Shilimkar wrote: > >> Mostly I was looking at the series from SOC boot and CPU PM >> point of view and boot part seems to just fine. >> >> As per discussion at LPC, I have gone through the SMC >> proposal which ARM has published. In general the boot part >> with SMC seems to be doable and can be standardized across SOCs. >> The part which will be conflicting is the CPU power management. >> That seems to be the harder one and the document is at too >> infancy stage from the details point of view. Some bits about >> save, restore are related to switcher kind of architecture, which >> may not be the requirement for all the SoCs. > > I've reviewed an earlier draft of that document. Although the examples > in the latest document appear to be geared towards switcher usage, my > suggestions to the ARM folks was to take into account the CPU hotplug > scenario instead. The provided examples remained switcher centric, but > the API in the published document is now much more generic than it used > to be. I think it should cover all usage scenarios now. > Thanks for the background Nicolas. Regards, Santosh -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH v3 00/31] AArch64 Linux kernel port
On Sat, 8 Sep 2012, Santosh Shilimkar wrote: > Mostly I was looking at the series from SOC boot and CPU PM > point of view and boot part seems to just fine. > > As per discussion at LPC, I have gone through the SMC > proposal which ARM has published. In general the boot part > with SMC seems to be doable and can be standardized across SOCs. > The part which will be conflicting is the CPU power management. > That seems to be the harder one and the document is at too > infancy stage from the details point of view. Some bits about > save, restore are related to switcher kind of architecture, which > may not be the requirement for all the SoCs. I've reviewed an earlier draft of that document. Although the examples in the latest document appear to be geared towards switcher usage, my suggestions to the ARM folks was to take into account the CPU hotplug scenario instead. The provided examples remained switcher centric, but the API in the published document is now much more generic than it used to be. I think it should cover all usage scenarios now. > Ofcourse the SMC discussion is not related to $subject series > as such. Also the PM support for ARMv8 not seems be the scope > of the series so the discussion can be parked aside till we > need to talk about it. If you wish to discuss this further (maybe in a separate thread) I'll be glad to discuss this with you. Nicolas -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH v3 00/31] AArch64 Linux kernel port
Catalin, On Friday 07 September 2012 09:56 PM, Catalin Marinas wrote: This is the 3rd version of patches implementing Linux kernel support for the 64-bit ARM architecture. Most of the review comments for the previous release have been implemented. There are a few left (like generic timer dts specification) but the main goal was to freeze the user ABI. The Linux kernel patches are available on this tree: Firstly thanks for clarifying the questions asked on the previous version and also for the update on the arch timer code. Mostly I was looking at the series from SOC boot and CPU PM point of view and boot part seems to just fine. As per discussion at LPC, I have gone through the SMC proposal which ARM has published. In general the boot part with SMC seems to be doable and can be standardized across SOCs. The part which will be conflicting is the CPU power management. That seems to be the harder one and the document is at too infancy stage from the details point of view. Some bits about save, restore are related to switcher kind of architecture, which may not be the requirement for all the SoCs. Ofcourse the SMC discussion is not related to $subject series as such. Also the PM support for ARMv8 not seems be the scope of the series so the discussion can be parked aside till we need to talk about it. I have few more questions on the various MM attribute usage, exceptions and PL's but I will do some reading before asking those to you. For the series itself, feel free to add my ack if yo need one. Acked-by: Santosh Shilimkar -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH v3 00/31] AArch64 Linux kernel port
Catalin, On Friday 07 September 2012 09:56 PM, Catalin Marinas wrote: This is the 3rd version of patches implementing Linux kernel support for the 64-bit ARM architecture. Most of the review comments for the previous release have been implemented. There are a few left (like generic timer dts specification) but the main goal was to freeze the user ABI. The Linux kernel patches are available on this tree: Firstly thanks for clarifying the questions asked on the previous version and also for the update on the arch timer code. Mostly I was looking at the series from SOC boot and CPU PM point of view and boot part seems to just fine. As per discussion at LPC, I have gone through the SMC proposal which ARM has published. In general the boot part with SMC seems to be doable and can be standardized across SOCs. The part which will be conflicting is the CPU power management. That seems to be the harder one and the document is at too infancy stage from the details point of view. Some bits about save, restore are related to switcher kind of architecture, which may not be the requirement for all the SoCs. Ofcourse the SMC discussion is not related to $subject series as such. Also the PM support for ARMv8 not seems be the scope of the series so the discussion can be parked aside till we need to talk about it. I have few more questions on the various MM attribute usage, exceptions and PL's but I will do some reading before asking those to you. For the series itself, feel free to add my ack if yo need one. Acked-by: Santosh Shilimkar santosh.shilim...@ti.com -- To unsubscribe from this list: send the line unsubscribe linux-kernel in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH v3 00/31] AArch64 Linux kernel port
On Sat, 8 Sep 2012, Santosh Shilimkar wrote: Mostly I was looking at the series from SOC boot and CPU PM point of view and boot part seems to just fine. As per discussion at LPC, I have gone through the SMC proposal which ARM has published. In general the boot part with SMC seems to be doable and can be standardized across SOCs. The part which will be conflicting is the CPU power management. That seems to be the harder one and the document is at too infancy stage from the details point of view. Some bits about save, restore are related to switcher kind of architecture, which may not be the requirement for all the SoCs. I've reviewed an earlier draft of that document. Although the examples in the latest document appear to be geared towards switcher usage, my suggestions to the ARM folks was to take into account the CPU hotplug scenario instead. The provided examples remained switcher centric, but the API in the published document is now much more generic than it used to be. I think it should cover all usage scenarios now. Ofcourse the SMC discussion is not related to $subject series as such. Also the PM support for ARMv8 not seems be the scope of the series so the discussion can be parked aside till we need to talk about it. If you wish to discuss this further (maybe in a separate thread) I'll be glad to discuss this with you. Nicolas -- To unsubscribe from this list: send the line unsubscribe linux-kernel in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH v3 00/31] AArch64 Linux kernel port
On Sat, Sep 8, 2012 at 7:29 PM, Nicolas Pitre n...@fluxnic.net wrote: On Sat, 8 Sep 2012, Santosh Shilimkar wrote: Mostly I was looking at the series from SOC boot and CPU PM point of view and boot part seems to just fine. As per discussion at LPC, I have gone through the SMC proposal which ARM has published. In general the boot part with SMC seems to be doable and can be standardized across SOCs. The part which will be conflicting is the CPU power management. That seems to be the harder one and the document is at too infancy stage from the details point of view. Some bits about save, restore are related to switcher kind of architecture, which may not be the requirement for all the SoCs. I've reviewed an earlier draft of that document. Although the examples in the latest document appear to be geared towards switcher usage, my suggestions to the ARM folks was to take into account the CPU hotplug scenario instead. The provided examples remained switcher centric, but the API in the published document is now much more generic than it used to be. I think it should cover all usage scenarios now. Thanks for the background Nicolas. Regards, Santosh -- To unsubscribe from this list: send the line unsubscribe linux-kernel in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH v3 00/31] AArch64 Linux kernel port
On Fri, Sep 07, 2012 at 05:26:35PM +0100, Catalin Marinas wrote: > This is the 3rd version of patches implementing Linux kernel support for > the 64-bit ARM architecture. Most of the review comments for the > previous release have been implemented. There are a few left (like > generic timer dts specification) but the main goal was to freeze the > user ABI. > > The Linux kernel patches are available on this tree: > > git://git.kernel.org/pub/scm/linux/kernel/git/cmarinas/linux-aarch64.git > upstream > > Main changes since the previous version: > - ptrace interface now using regsets for TLS and hardware debug > registers. > - sys_personality fixes. PER_LINUX is the default for all tasks (32 or > 64-bit) and there is no compat_sys_personality defined. > - Kconfig clean-up (unused symbols, AARCH32_EMULATION removed in favour > of COMPAT). > - Clean-up of the __cpuc_* function prefixes. > - CPU information table implemented in C rather than assembly. > - __bad_* function calls replaced with BUILD_BUG. > - Inline get_user/put_user (code cleaner, slightly bigger kernel Image). > - DMA API implementation simplified, defaulting to swiotlb. It will be > extended as hardware becomes available. > - Arch-specific code in the generic timer patch moved to arch/arm64/. > - Comment fixes and clarifications. I see that a good amount of the comments I had on v2 have neither been replied to nor addressed. Please address them where appropriate, ideally before merging. I'll be happy to be told I'm wrong in some of the cases too, but completely ignoring feedback isn't cool. However, most of them were nits, and I've had a look-through of the rest of the patches, feel free to add: Acked-by: Olof Johansson on the patches in this series. -Olof -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH v3 00/31] AArch64 Linux kernel port
On Fri, Sep 07, 2012 at 05:26:35PM +0100, Catalin Marinas wrote: This is the 3rd version of patches implementing Linux kernel support for the 64-bit ARM architecture. Most of the review comments for the previous release have been implemented. There are a few left (like generic timer dts specification) but the main goal was to freeze the user ABI. The Linux kernel patches are available on this tree: git://git.kernel.org/pub/scm/linux/kernel/git/cmarinas/linux-aarch64.git upstream Main changes since the previous version: - ptrace interface now using regsets for TLS and hardware debug registers. - sys_personality fixes. PER_LINUX is the default for all tasks (32 or 64-bit) and there is no compat_sys_personality defined. - Kconfig clean-up (unused symbols, AARCH32_EMULATION removed in favour of COMPAT). - Clean-up of the __cpuc_* function prefixes. - CPU information table implemented in C rather than assembly. - __bad_* function calls replaced with BUILD_BUG. - Inline get_user/put_user (code cleaner, slightly bigger kernel Image). - DMA API implementation simplified, defaulting to swiotlb. It will be extended as hardware becomes available. - Arch-specific code in the generic timer patch moved to arch/arm64/. - Comment fixes and clarifications. I see that a good amount of the comments I had on v2 have neither been replied to nor addressed. Please address them where appropriate, ideally before merging. I'll be happy to be told I'm wrong in some of the cases too, but completely ignoring feedback isn't cool. However, most of them were nits, and I've had a look-through of the rest of the patches, feel free to add: Acked-by: Olof Johansson o...@lixom.net on the patches in this series. -Olof -- To unsubscribe from this list: send the line unsubscribe linux-kernel in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/