| Message ID | 20230106220020.1820147-1-anirudh.venkataramanan@intel.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | Remove three Sun net drivers | expand |
On 1/6/23 15:00, Anirudh Venkataramanan wrote: > This series removes the Sun Cassini, LDOM vswitch and sunvnet drivers. > > In a recent patch series that touched these drivers [1], it was suggested > that these drivers should be removed completely. git logs suggest that > there hasn't been any significant feature addition, improvement or fixes > to user-visible bugs in a while. A web search didn't indicate any recent > discussions or any evidence that there are users out there who care about > these drivers. > > The idea behind putting out this series is to either establish that these > drivers are used and should be maintained, or remove them. Anirudh, The Sun LDOM vswitch and sunvnet drivers are still in use, please do not remove them or the event tracing. We use them internally and you don't see any discussions because they generally work fine (AFAIK). I think you are also going to break things by removing Sun Cassini support, but I am not using it personally. What user visible bugs are you referring to here? Thanks Karl > > While the bulk of the code removal is in the networking s, and so > multiple subsystem lists are cc'd. Here's a quick breakdown: > > - patches 1/7 and 5/7 remove the drivers (netdev) > - patch 2/7 removes Sun device IDs from pci_ids.h (linux-pci) > - patch 3/7 changes ppc6xx_defconfig (linuxppc) > - patch 4/7 changes MIPS mtx1_defconfig (linux-mips) > - patch 6/7 removes the event tracing header sunvnet.h (linux-trace) > - patch 7/7 changes sparc64_defconfig (sparclinux) > > This series was compile tested as follows: > > make O=b1 ARCH=mips CROSS_COMPILE=mips64-linux-gnu- defconfig > make -j$(nproc) O=b1 ARCH=mips CROSS_COMPILE=mips64-linux-gnu- all > > make O=b2 ARCH=sparc64 CROSS_COMPILE=sparc64-linux-gnu- defconfig > make -j$(nproc) O=b2 ARCH=sparc64 CROSS_COMPILE=sparc64-linux-gnu- all > > make O=b3 ARCH=powerpc CROSS_COMPILE=ppc64-linux-gnu- ppc6xx_defconfig > make -j$(nproc) O=b3 ARCH=powerpc CROSS_COMPILE=ppc64-linux-gnu- all > > [1] https://lore.kernel.org/netdev/99629223-ac1b-0f82-50b8-ea307b3b0197@intel.com/T/#t > > Suggested-by: Leon Romanovsky <leon@kernel.org> > Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> > > Anirudh Venkataramanan (7): > ethernet: Remove the Sun Cassini driver > PCI: Remove PCI IDs used by the Sun Cassini driver > powerpc: configs: Remove reference to CONFIG_CASSINI > mips: configs: Remove reference to CONFIG_CASSINI > ethernet: Remove the Sun LDOM vswitch and sunvnet drivers > sunvnet: Remove event tracing file > sparc: configs: Remove references to CONFIG_SUNVNET and CONFIG_LDMVSW > > arch/mips/configs/mtx1_defconfig | 1 - > arch/powerpc/configs/ppc6xx_defconfig | 1 - > arch/sparc/configs/sparc64_defconfig | 2 - > drivers/net/ethernet/sun/Kconfig | 35 - > drivers/net/ethernet/sun/Makefile | 4 - > drivers/net/ethernet/sun/cassini.c | 5215 --------------------- > drivers/net/ethernet/sun/cassini.h | 2900 ------------ > drivers/net/ethernet/sun/ldmvsw.c | 476 -- > drivers/net/ethernet/sun/sunvnet.c | 567 --- > drivers/net/ethernet/sun/sunvnet_common.c | 1813 ------- > drivers/net/ethernet/sun/sunvnet_common.h | 157 - > include/linux/pci_ids.h | 2 - > include/trace/events/sunvnet.h | 140 - > 13 files changed, 11313 deletions(-) > delete mode 100644 drivers/net/ethernet/sun/cassini.c > delete mode 100644 drivers/net/ethernet/sun/cassini.h > delete mode 100644 drivers/net/ethernet/sun/ldmvsw.c > delete mode 100644 drivers/net/ethernet/sun/sunvnet.c > delete mode 100644 drivers/net/ethernet/sun/sunvnet_common.c > delete mode 100644 drivers/net/ethernet/sun/sunvnet_common.h > delete mode 100644 include/trace/events/sunvnet.h > > > base-commit: 6bd4755c7c499dbcef46eaaeafa1a319da583b29
On 1/6/2023 2:44 PM, Karl Volz wrote: > > > On 1/6/23 15:00, Anirudh Venkataramanan wrote: >> This series removes the Sun Cassini, LDOM vswitch and sunvnet drivers. >> >> In a recent patch series that touched these drivers [1], it was suggested >> that these drivers should be removed completely. git logs suggest that >> there hasn't been any significant feature addition, improvement or fixes >> to user-visible bugs in a while. A web search didn't indicate any recent >> discussions or any evidence that there are users out there who care about >> these drivers. >> >> The idea behind putting out this series is to either establish that these >> drivers are used and should be maintained, or remove them. > Anirudh, > > The Sun LDOM vswitch and sunvnet drivers are still in use, please do not > remove them or the event tracing. > We use them internally and you don't see any discussions because they > generally work fine (AFAIK). Hello Karl, Thanks for chiming in. Are there recent platforms where these drivers are used? If yes, do you know which ones? Or are these drivers useful in old/legacy platforms that are still around but perhaps no longer in production? > I think you are also going to break things by removing Sun Cassini > support, but I am not using it personally. You suspect there are users for this driver as well? > What user visible bugs are you referring to here? I was saying I don't see any evidence of recent bug fixes, which would make sense if these drivers "just work". Ani
On 1/6/23 16:10, Anirudh Venkataramanan wrote: > On 1/6/2023 2:44 PM, Karl Volz wrote: >> >> >> On 1/6/23 15:00, Anirudh Venkataramanan wrote: >>> This series removes the Sun Cassini, LDOM vswitch and sunvnet drivers. >>> >>> In a recent patch series that touched these drivers [1], it was >>> suggested >>> that these drivers should be removed completely. git logs suggest that >>> there hasn't been any significant feature addition, improvement or >>> fixes >>> to user-visible bugs in a while. A web search didn't indicate any >>> recent >>> discussions or any evidence that there are users out there who care >>> about >>> these drivers. >>> >>> The idea behind putting out this series is to either establish that >>> these >>> drivers are used and should be maintained, or remove them. >> Anirudh, >> >> The Sun LDOM vswitch and sunvnet drivers are still in use, please do >> not remove them or the event tracing. >> We use them internally and you don't see any discussions because they >> generally work fine (AFAIK). > > Hello Karl, > > Thanks for chiming in. > > Are there recent platforms where these drivers are used? If yes, do > you know which ones? Or are these drivers useful in old/legacy > platforms that are still around but perhaps no longer in production? These drivers work on older T4, T5, etc to the latest T7, T8 Sun servers (e.g. T8-2, T8-4, note, T7/T8 are still in production). They may also work on T2/T3 (but I don't use those anymore, though Adrian (John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>) might). You might have missed a lot of the linux SPARC developers since they are on the debian-sparc@lists.debian.org list, hence I'd ask there also. > >> I think you are also going to break things by removing Sun Cassini >> support, but I am not using it personally. > > You suspect there are users for this driver as well? Yes, they may not have seen this yet or recognized the servers this card goes in (post on debian-sparc@lists.debian.org list). > >> What user visible bugs are you referring to here? > > I was saying I don't see any evidence of recent bug fixes, which would > make sense if these drivers "just work". ok, no worries. Cheers Karl > > Ani
Hello! On 1/6/23 23:00, Anirudh Venkataramanan wrote: > This series removes the Sun Cassini, LDOM vswitch and sunvnet drivers. This would affect a large number of Linux on SPARC users. Please don't! We're still maintaining an active sparc64 port for Debian, see [1]. So does Gentoo [2]. > In a recent patch series that touched these drivers [1], it was suggested > that these drivers should be removed completely. git logs suggest that > there hasn't been any significant feature addition, improvement or fixes > to user-visible bugs in a while. A web search didn't indicate any recent > discussions or any evidence that there are users out there who care about > these drivers. Well, these drivers just work and I don't see why there should be regular discussions about them or changes. Adrian > [1] https://cdimage.debian.org/cdimage/ports/snapshots/2022-12-09/ > [2] https://www.gentoo.org/downloads/
On 1/6/2023 5:36 PM, John Paul Adrian Glaubitz wrote: > Hello! > > On 1/6/23 23:00, Anirudh Venkataramanan wrote: >> This series removes the Sun Cassini, LDOM vswitch and sunvnet drivers. > > This would affect a large number of Linux on SPARC users. Please don't! Thanks for chiming in. Does your statement above apply to all 3 drivers? > > We're still maintaining an active sparc64 port for Debian, see [1]. So > does Gentoo [2]. > >> In a recent patch series that touched these drivers [1], it was suggested >> that these drivers should be removed completely. git logs suggest that >> there hasn't been any significant feature addition, improvement or fixes >> to user-visible bugs in a while. A web search didn't indicate any recent >> discussions or any evidence that there are users out there who care about >> these drivers. > > Well, these drivers just work and I don't see why there should be regular > discussions about them or changes. That's fair, but lack of discussion can also be signs of disuse, and that's really the hunch I was following up on. Given what you and Karl have said, I agree that we shouldn't remove these drivers. I'll stop pursuing this unless there are new arguments to the contrary. Ani
On Fri, 6 Jan 2023 18:04:54 -0800 Anirudh Venkataramanan wrote: > >> In a recent patch series that touched these drivers [1], it was suggested > >> that these drivers should be removed completely. git logs suggest that > >> there hasn't been any significant feature addition, improvement or fixes > >> to user-visible bugs in a while. A web search didn't indicate any recent > >> discussions or any evidence that there are users out there who care about > >> these drivers. > > > > Well, these drivers just work and I don't see why there should be regular > > discussions about them or changes. > > That's fair, but lack of discussion can also be signs of disuse, and > that's really the hunch I was following up on. Lack of feedback, too. Some of these drivers are missing entries in MAINTAINERS and patches don't get review tags from anyone.
On 1/7/23 03:04, Anirudh Venkataramanan wrote: > On 1/6/2023 5:36 PM, John Paul Adrian Glaubitz wrote: >> Hello! >> >> On 1/6/23 23:00, Anirudh Venkataramanan wrote: >>> This series removes the Sun Cassini, LDOM vswitch and sunvnet drivers. >> >> This would affect a large number of Linux on SPARC users. Please don't! > > Thanks for chiming in. Does your statement above apply to all 3 drivers? Yes! >> We're still maintaining an active sparc64 port for Debian, see [1]. So >> does Gentoo [2]. >> >>> In a recent patch series that touched these drivers [1], it was suggested >>> that these drivers should be removed completely. git logs suggest that >>> there hasn't been any significant feature addition, improvement or fixes >>> to user-visible bugs in a while. A web search didn't indicate any recent >>> discussions or any evidence that there are users out there who care about >>> these drivers. >> >> Well, these drivers just work and I don't see why there should be regular >> discussions about them or changes. > > That's fair, but lack of discussion can also be signs of disuse, and that's > really the hunch I was following up on. Given what you and Karl have said, > I agree that we shouldn't remove these drivers. I'll stop pursuing this unless > there are new arguments to the contrary. It's a common problem in my opinion on the LKML that many kernel developers assume that users of certain drivers and kernel subsystems are present and active on the kernel mailing lists to be able to raise their voices in these discussions. If you want to find out whether some parts of the kernel are actively being used, it's better to ask on distribution mailing lists because it's way more likely to find any users there. I try to be present on as many kernel mailing lists as I can to be able to answer these questions, but sometimes there is just too much traffic for me to handle. Adrian
On Sat, Jan 7, 2023 at 1:00 AM Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> wrote: > > In a recent patch series that touched this driver [1], it was suggested > that this driver should be removed completely. git logs suggest that > there hasn't been any significant feature addition, improvement or fixes to > user-visible bugs in a while. A web search didn't indicate any recent > discussions or any evidence that there are users out there who care about > this driver. Thus, remove this driver. > > Notes: > > checkpatch complains "WARNING: added, moved or deleted file(s), does > MAINTAINERS need updating?". The files being removed don't have their > own entries in the MAINTAINERS file, so there's nothing to remove. > > checkpatch also complains about the long lore link below. > > [1] https://lore.kernel.org/netdev/99629223-ac1b-0f82-50b8-ea307b3b0197@intel.com/T/#t > > Suggested-by: Leon Romanovsky <leon@kernel.org> > Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Do we drop/delete a working functionality by only taking in account git activity ? What is a proper way to decline patch series (vs Acked-by) ? Thanks.
On 1/7/2023 4:25 AM, Anatoly Pugachev wrote: > On Sat, Jan 7, 2023 at 1:00 AM Anirudh Venkataramanan > <anirudh.venkataramanan@intel.com> wrote: >> >> In a recent patch series that touched this driver [1], it was suggested >> that this driver should be removed completely. git logs suggest that >> there hasn't been any significant feature addition, improvement or fixes to >> user-visible bugs in a while. A web search didn't indicate any recent >> discussions or any evidence that there are users out there who care about >> this driver. Thus, remove this driver. >> >> Notes: >> >> checkpatch complains "WARNING: added, moved or deleted file(s), does >> MAINTAINERS need updating?". The files being removed don't have their >> own entries in the MAINTAINERS file, so there's nothing to remove. >> >> checkpatch also complains about the long lore link below. >> >> [1] https://lore.kernel.org/netdev/99629223-ac1b-0f82-50b8-ea307b3b0197@intel.com/T/#t >> >> Suggested-by: Leon Romanovsky <leon@kernel.org> >> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> > > Do we drop/delete a working functionality by only taking in account > git activity ? No, but in some cases it's enough to at least start asking the "who uses this code? should we continue maintaining it?" type questions. In the cover letter I did say this: "The idea behind putting out this series is to either establish that these drivers are used and should be maintained, or remove them." We have established that these drivers are indeed used, and thus shouldn't be removed. > > What is a proper way to decline patch series (vs Acked-by) ? There's no tag that I am aware of. I have seen people say "NACK" or "please don't do this" followed by an explanation of why the patch/series is a bad idea. For example, see the other responses to this series. Ani
This series removes the Sun Cassini, LDOM vswitch and sunvnet drivers. In a recent patch series that touched these drivers [1], it was suggested that these drivers should be removed completely. git logs suggest that there hasn't been any significant feature addition, improvement or fixes to user-visible bugs in a while. A web search didn't indicate any recent discussions or any evidence that there are users out there who care about these drivers. The idea behind putting out this series is to either establish that these drivers are used and should be maintained, or remove them. While the bulk of the code removal is in the networking s, and so multiple subsystem lists are cc'd. Here's a quick breakdown: - patches 1/7 and 5/7 remove the drivers (netdev) - patch 2/7 removes Sun device IDs from pci_ids.h (linux-pci) - patch 3/7 changes ppc6xx_defconfig (linuxppc) - patch 4/7 changes MIPS mtx1_defconfig (linux-mips) - patch 6/7 removes the event tracing header sunvnet.h (linux-trace) - patch 7/7 changes sparc64_defconfig (sparclinux) This series was compile tested as follows: make O=b1 ARCH=mips CROSS_COMPILE=mips64-linux-gnu- defconfig make -j$(nproc) O=b1 ARCH=mips CROSS_COMPILE=mips64-linux-gnu- all make O=b2 ARCH=sparc64 CROSS_COMPILE=sparc64-linux-gnu- defconfig make -j$(nproc) O=b2 ARCH=sparc64 CROSS_COMPILE=sparc64-linux-gnu- all make O=b3 ARCH=powerpc CROSS_COMPILE=ppc64-linux-gnu- ppc6xx_defconfig make -j$(nproc) O=b3 ARCH=powerpc CROSS_COMPILE=ppc64-linux-gnu- all [1] https://lore.kernel.org/netdev/99629223-ac1b-0f82-50b8-ea307b3b0197@intel.com/T/#t Suggested-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com> Anirudh Venkataramanan (7): ethernet: Remove the Sun Cassini driver PCI: Remove PCI IDs used by the Sun Cassini driver powerpc: configs: Remove reference to CONFIG_CASSINI mips: configs: Remove reference to CONFIG_CASSINI ethernet: Remove the Sun LDOM vswitch and sunvnet drivers sunvnet: Remove event tracing file sparc: configs: Remove references to CONFIG_SUNVNET and CONFIG_LDMVSW arch/mips/configs/mtx1_defconfig | 1 - arch/powerpc/configs/ppc6xx_defconfig | 1 - arch/sparc/configs/sparc64_defconfig | 2 - drivers/net/ethernet/sun/Kconfig | 35 - drivers/net/ethernet/sun/Makefile | 4 - drivers/net/ethernet/sun/cassini.c | 5215 --------------------- drivers/net/ethernet/sun/cassini.h | 2900 ------------ drivers/net/ethernet/sun/ldmvsw.c | 476 -- drivers/net/ethernet/sun/sunvnet.c | 567 --- drivers/net/ethernet/sun/sunvnet_common.c | 1813 ------- drivers/net/ethernet/sun/sunvnet_common.h | 157 - include/linux/pci_ids.h | 2 - include/trace/events/sunvnet.h | 140 - 13 files changed, 11313 deletions(-) delete mode 100644 drivers/net/ethernet/sun/cassini.c delete mode 100644 drivers/net/ethernet/sun/cassini.h delete mode 100644 drivers/net/ethernet/sun/ldmvsw.c delete mode 100644 drivers/net/ethernet/sun/sunvnet.c delete mode 100644 drivers/net/ethernet/sun/sunvnet_common.c delete mode 100644 drivers/net/ethernet/sun/sunvnet_common.h delete mode 100644 include/trace/events/sunvnet.h base-commit: 6bd4755c7c499dbcef46eaaeafa1a319da583b29