| Message ID | 20211007000954.30621-1-zev@bewilderbeest.net (mailing list archive) |
|---|---|
| Headers | show |
| Series | Dynamic DT device nodes | expand |
On 10/6/2021 5:09 PM, Zev Weiss wrote: > Hello, > > This patch series is in some ways kind of a v2 for the "Dynamic > aspeed-smc flash chips via 'reserved' DT status" series I posted > previously [0], but takes a fairly different approach suggested by Rob > Herring [1] and doesn't actually touch the aspeed-smc driver or > anything in the MTD subsystem, so I haven't marked it as such. > > To recap a bit of the context from that series, in OpenBMC there's a > need for certain devices (described by device-tree nodes) to be able > to be attached and detached at runtime (for example the SPI flash for > the host's firmware, which is shared between the BMC and the host but > can only be accessed by one or the other at a time). To provide that > ability, this series adds support for a new common device-tree > property, a boolean "dynamic" that indicates that the device may come > and go at runtime. When present on a node, the sysfs file for that > node's "status" property is made writable, allowing userspace to do > things like: > > $ echo okay > /sys/firmware/devicetree/.../status > $ echo reserved > /sys/firmware/devicetree/.../status > > to activate and deactivate a dynamic device. This is a completely drive by comment here, but cannot you already achieve what you want today by making the SPI-NOR to be loaded as a module, probe it when you need it from the BMC, and unbind or rmmod the drive when you need it on the server/host attached to the BMC? Looking at [0] there appears to be enough signaling visible by the BMC's user-space that it ought to be possible? Assuming that there may be multiple flash chips and you somehow need to access one in order to complete the BMC device boot, but not the other one(s), you could imagine unbinding the spi-nor driver from the ones you don't want to drive and wait until you have appropriate signaling made available to your or is there a risk of racing with the host in doing so?
On Wed, Oct 06, 2021 at 07:46:08PM PDT, Florian Fainelli wrote: > > >On 10/6/2021 5:09 PM, Zev Weiss wrote: >>Hello, >> >>This patch series is in some ways kind of a v2 for the "Dynamic >>aspeed-smc flash chips via 'reserved' DT status" series I posted >>previously [0], but takes a fairly different approach suggested by Rob >>Herring [1] and doesn't actually touch the aspeed-smc driver or >>anything in the MTD subsystem, so I haven't marked it as such. >> >>To recap a bit of the context from that series, in OpenBMC there's a >>need for certain devices (described by device-tree nodes) to be able >>to be attached and detached at runtime (for example the SPI flash for >>the host's firmware, which is shared between the BMC and the host but >>can only be accessed by one or the other at a time). To provide that >>ability, this series adds support for a new common device-tree >>property, a boolean "dynamic" that indicates that the device may come >>and go at runtime. When present on a node, the sysfs file for that >>node's "status" property is made writable, allowing userspace to do >>things like: >> >> $ echo okay > /sys/firmware/devicetree/.../status >> $ echo reserved > /sys/firmware/devicetree/.../status >> >>to activate and deactivate a dynamic device. > >This is a completely drive by comment here, but cannot you already >achieve what you want today by making the SPI-NOR to be loaded as a >module, probe it when you need it from the BMC, and unbind or rmmod >the drive when you need it on the server/host attached to the BMC? > >Looking at [0] there appears to be enough signaling visible by the >BMC's user-space that it ought to be possible? > >Assuming that there may be multiple flash chips and you somehow need >to access one in order to complete the BMC device boot, but not the >other one(s), you could imagine unbinding the spi-nor driver from the >ones you don't want to drive and wait until you have appropriate >signaling made available to your or is there a risk of racing with the >host in doing so? Hi Florian, I sort of considered similar things at various points (in fact the bind/unbind technique is what I've been using as a stopgap so far), but I don't think it's ultimately a great solution. In this particular case it happens that the driver for the host's BIOS flash (aspeed-smc) is the same driver used for the BMC's firmware flash and hence necessary for it to boot. (OpenBMC also typically uses a non-modular kernel, for what it's worth.) On BMC startup we don't know the state of the host, and while the mux that ultimately determines which processor has access to the BIOS flash defaults to connecting it to the host (so if we do attempt to attach it we'll simply fail clumsily and spew some errors to dmesg rather than causing catastrophic errors host-side), blindly starting to poke the BIOS flash without doing the proper coordination with the host first seems to me to be pretty squarely in the category of "things we shouldn't be doing" -- the desire to avoid that sort of clunkiness is a significant portion of what led me to pursue a better solution in the first place. Additionally, while I don't know the details of the specific hardware involved, others in the OpenBMC community (or at least Andrew Jeffery) have indicated that pluggable DT devices would be useful for the systems they're working on as well, I suspect for things other than host firmware flashes. Zev
On Thu, Oct 7, 2021 at 3:10 AM Zev Weiss <zev@bewilderbeest.net> wrote: > This patch series is in some ways kind of a v2 for the "Dynamic > aspeed-smc flash chips via 'reserved' DT status" series I posted > previously [0], but takes a fairly different approach suggested by Rob > Herring [1] and doesn't actually touch the aspeed-smc driver or > anything in the MTD subsystem, so I haven't marked it as such. > > To recap a bit of the context from that series, in OpenBMC there's a > need for certain devices (described by device-tree nodes) to be able > to be attached and detached at runtime (for example the SPI flash for > the host's firmware, which is shared between the BMC and the host but > can only be accessed by one or the other at a time). This seems quite dangerous. Why do you need that? Why can't device tree overlays be used? > To provide that > ability, this series adds support for a new common device-tree > property, a boolean "dynamic" that indicates that the device may come > and go at runtime. When present on a node, the sysfs file for that > node's "status" property is made writable, allowing userspace to do > things like: > > $ echo okay > /sys/firmware/devicetree/.../status > $ echo reserved > /sys/firmware/devicetree/.../status > > to activate and deactivate a dynamic device. > > Because it leans on the OF_DYNAMIC machinery internally, this > functionality will only work on busses that register for OF reconfig > notifications and handle them appropriately (presently platform, i2c, > and spi). This series does not attempt to solve the "dynamic devices > further down the tree" problem [2]; my hope is that handling for OF > reconfig notifications can be extended to other families of devices > (e.g. individual MTD spi-nor flash chips) in the future. What about ACPI and software nodes? How will all this affect the host? > [0] https://lore.kernel.org/openbmc/20210929115409.21254-1-zev@bewilderbeest.net/ > [1] https://lore.kernel.org/openbmc/CAL_JsqJH+b5oFuSP+KBLBsN5QTA6xASuqXJWXUaDkHhugXPpnQ@mail.gmail.com/ > [2] https://lore.kernel.org/openbmc/20210929220038.GS17315@packtop/ > [3] https://git.kernel.org/pub/scm/linux/kernel/git/robh/linux.git/commit/?id=6663ae07d995f5fbe2988a19858b2f87e68cf929
On Thu, Oct 07, 2021 at 12:04:41AM PDT, Andy Shevchenko wrote: >On Thu, Oct 7, 2021 at 3:10 AM Zev Weiss <zev@bewilderbeest.net> wrote: >> This patch series is in some ways kind of a v2 for the "Dynamic >> aspeed-smc flash chips via 'reserved' DT status" series I posted >> previously [0], but takes a fairly different approach suggested by Rob >> Herring [1] and doesn't actually touch the aspeed-smc driver or >> anything in the MTD subsystem, so I haven't marked it as such. >> >> To recap a bit of the context from that series, in OpenBMC there's a >> need for certain devices (described by device-tree nodes) to be able >> to be attached and detached at runtime (for example the SPI flash for >> the host's firmware, which is shared between the BMC and the host but >> can only be accessed by one or the other at a time). > >This seems quite dangerous. Why do you need that? Sometimes the host needs access to the flash (it's the host's firmware, after all), sometimes the BMC needs access to it (e.g. to perform an out-of-band update to the host's firmware). To achieve the latter, the flash needs to be attached to the BMC, but that requires some careful coordination with the host to arbitrate which one actually has access to it (that coordination is handled by userspace, which then tells the kernel explicitly when the flash should be attached and detached). What seems dangerous? >Why can't device tree overlays be used? I'm hoping to stay closer to mainline. The OpenBMC kernel has a documented policy strongly encouraging upstream-first development: https://github.com/openbmc/docs/blob/master/kernel-development.md I doubt Joel (the OpenBMC kernel maintainer) would be eager to start carrying the DT overlay patches; I'd likewise strongly prefer to avoid carrying them myself as additional downstream patches. Hence the attempt at getting a solution to the problem upstream. > >> To provide that >> ability, this series adds support for a new common device-tree >> property, a boolean "dynamic" that indicates that the device may come >> and go at runtime. When present on a node, the sysfs file for that >> node's "status" property is made writable, allowing userspace to do >> things like: >> >> $ echo okay > /sys/firmware/devicetree/.../status >> $ echo reserved > /sys/firmware/devicetree/.../status >> >> to activate and deactivate a dynamic device. >> >> Because it leans on the OF_DYNAMIC machinery internally, this >> functionality will only work on busses that register for OF reconfig >> notifications and handle them appropriately (presently platform, i2c, >> and spi). This series does not attempt to solve the "dynamic devices >> further down the tree" problem [2]; my hope is that handling for OF >> reconfig notifications can be extended to other families of devices >> (e.g. individual MTD spi-nor flash chips) in the future. > >What about ACPI and software nodes? I'm afraid I don't understand the question, can you elaborate on what you mean? >How will all this affect the host? Assuming the coordination mentioned above is done properly, the host will be in a quiesced state whenever the BMC is accessing the flash and hence won't notice much of anything at all (the BMC will detach the flash and relinquish control of it back to the host before the host is reactivated). Zev
On Thu, Oct 07, 2021 at 02:05:41AM -0700, Zev Weiss wrote: > On Thu, Oct 07, 2021 at 12:04:41AM PDT, Andy Shevchenko wrote: > > On Thu, Oct 7, 2021 at 3:10 AM Zev Weiss <zev@bewilderbeest.net> wrote: > > > This patch series is in some ways kind of a v2 for the "Dynamic > > > aspeed-smc flash chips via 'reserved' DT status" series I posted > > > previously [0], but takes a fairly different approach suggested by Rob > > > Herring [1] and doesn't actually touch the aspeed-smc driver or > > > anything in the MTD subsystem, so I haven't marked it as such. > > > > > > To recap a bit of the context from that series, in OpenBMC there's a > > > need for certain devices (described by device-tree nodes) to be able > > > to be attached and detached at runtime (for example the SPI flash for > > > the host's firmware, which is shared between the BMC and the host but > > > can only be accessed by one or the other at a time). > > > > This seems quite dangerous. Why do you need that? > > Sometimes the host needs access to the flash (it's the host's firmware, > after all), sometimes the BMC needs access to it (e.g. to perform an > out-of-band update to the host's firmware). To achieve the latter, the > flash needs to be attached to the BMC, but that requires some careful > coordination with the host to arbitrate which one actually has access to it > (that coordination is handled by userspace, which then tells the kernel > explicitly when the flash should be attached and detached). > > What seems dangerous? > > > Why can't device tree overlays be used? > > I'm hoping to stay closer to mainline. The OpenBMC kernel has a documented > policy strongly encouraging upstream-first development: > https://github.com/openbmc/docs/blob/master/kernel-development.md > > I doubt Joel (the OpenBMC kernel maintainer) would be eager to start > carrying the DT overlay patches; I'd likewise strongly prefer to avoid > carrying them myself as additional downstream patches. Hence the attempt at > getting a solution to the problem upstream. Then why not work to get device tree overlays to be merged properly? Don't work on a half-of-a-solution when the real solution is already here. thanks, greg k-h
On Thu, Oct 07, 2021 at 03:31:39AM PDT, Greg Kroah-Hartman wrote: >On Thu, Oct 07, 2021 at 02:05:41AM -0700, Zev Weiss wrote: >> On Thu, Oct 07, 2021 at 12:04:41AM PDT, Andy Shevchenko wrote: >> > On Thu, Oct 7, 2021 at 3:10 AM Zev Weiss <zev@bewilderbeest.net> wrote: >> > > This patch series is in some ways kind of a v2 for the "Dynamic >> > > aspeed-smc flash chips via 'reserved' DT status" series I posted >> > > previously [0], but takes a fairly different approach suggested by Rob >> > > Herring [1] and doesn't actually touch the aspeed-smc driver or >> > > anything in the MTD subsystem, so I haven't marked it as such. >> > > >> > > To recap a bit of the context from that series, in OpenBMC there's a >> > > need for certain devices (described by device-tree nodes) to be able >> > > to be attached and detached at runtime (for example the SPI flash for >> > > the host's firmware, which is shared between the BMC and the host but >> > > can only be accessed by one or the other at a time). >> > >> > This seems quite dangerous. Why do you need that? >> >> Sometimes the host needs access to the flash (it's the host's firmware, >> after all), sometimes the BMC needs access to it (e.g. to perform an >> out-of-band update to the host's firmware). To achieve the latter, the >> flash needs to be attached to the BMC, but that requires some careful >> coordination with the host to arbitrate which one actually has access to it >> (that coordination is handled by userspace, which then tells the kernel >> explicitly when the flash should be attached and detached). >> >> What seems dangerous? >> >> > Why can't device tree overlays be used? >> >> I'm hoping to stay closer to mainline. The OpenBMC kernel has a documented >> policy strongly encouraging upstream-first development: >> https://github.com/openbmc/docs/blob/master/kernel-development.md >> >> I doubt Joel (the OpenBMC kernel maintainer) would be eager to start >> carrying the DT overlay patches; I'd likewise strongly prefer to avoid >> carrying them myself as additional downstream patches. Hence the attempt at >> getting a solution to the problem upstream. > >Then why not work to get device tree overlays to be merged properly? >Don't work on a half-of-a-solution when the real solution is already >here. > I had been under the impression that the overlay patches had very dim prospects of ever being accepted and that this might be a more tractable alternative, but apparently was mistaken -- I'll look into what the outstanding issues were with that and perhaps take a stab at addressing them. Zev
On Thu, Oct 7, 2021 at 10:41 AM Zev Weiss <zev@bewilderbeest.net> wrote: > > On Thu, Oct 07, 2021 at 03:31:39AM PDT, Greg Kroah-Hartman wrote: > >On Thu, Oct 07, 2021 at 02:05:41AM -0700, Zev Weiss wrote: > >> On Thu, Oct 07, 2021 at 12:04:41AM PDT, Andy Shevchenko wrote: > >> > On Thu, Oct 7, 2021 at 3:10 AM Zev Weiss <zev@bewilderbeest.net> wrote: > >> > > This patch series is in some ways kind of a v2 for the "Dynamic > >> > > aspeed-smc flash chips via 'reserved' DT status" series I posted > >> > > previously [0], but takes a fairly different approach suggested by Rob > >> > > Herring [1] and doesn't actually touch the aspeed-smc driver or > >> > > anything in the MTD subsystem, so I haven't marked it as such. > >> > > > >> > > To recap a bit of the context from that series, in OpenBMC there's a > >> > > need for certain devices (described by device-tree nodes) to be able > >> > > to be attached and detached at runtime (for example the SPI flash for > >> > > the host's firmware, which is shared between the BMC and the host but > >> > > can only be accessed by one or the other at a time). > >> > > >> > This seems quite dangerous. Why do you need that? > >> > >> Sometimes the host needs access to the flash (it's the host's firmware, > >> after all), sometimes the BMC needs access to it (e.g. to perform an > >> out-of-band update to the host's firmware). To achieve the latter, the > >> flash needs to be attached to the BMC, but that requires some careful > >> coordination with the host to arbitrate which one actually has access to it > >> (that coordination is handled by userspace, which then tells the kernel > >> explicitly when the flash should be attached and detached). > >> > >> What seems dangerous? > >> > >> > Why can't device tree overlays be used? > >> > >> I'm hoping to stay closer to mainline. The OpenBMC kernel has a documented > >> policy strongly encouraging upstream-first development: > >> https://github.com/openbmc/docs/blob/master/kernel-development.md > >> > >> I doubt Joel (the OpenBMC kernel maintainer) would be eager to start > >> carrying the DT overlay patches; I'd likewise strongly prefer to avoid > >> carrying them myself as additional downstream patches. Hence the attempt at > >> getting a solution to the problem upstream. > > > >Then why not work to get device tree overlays to be merged properly? TBC, it's 'just' the general purpose userspace interface to apply overlays that's missing. I did suggest what's done here as overlays are kind of an overkill for this usecase. Much easier to write to a sysfs file than write an overlay, compile it with dtc, and provide it to the kernel all just to enable a device. Perhaps this could also be supported in the driver model directly. Given the "what about ACPI question", that is probably what should be done unless the answer is we don't care. I think we'd just need a flag to create devices, but not bind automatically. Or maybe abusing driver_override can accomplish that. > >Don't work on a half-of-a-solution when the real solution is already > >here. > > > > I had been under the impression that the overlay patches had very dim > prospects of ever being accepted and that this might be a more tractable > alternative, but apparently was mistaken -- I'll look into what the > outstanding issues were with that and perhaps take a stab at addressing > them. What's dim is the patches allowing any modification to any part of the DT. Any changes to a DT need to work (i.e. have some effect). For example, randomly changing/adding/removing properties wouldn't have any effect because they've probably already be read and used. What I've suggested before is some sort of registration of nodes allowed to apply child nodes and properties to. That would serve the add-on board usecases which have been the main driver of this to date. That also got hung up on defining interface nodes to add-on boards. Your scope is more limited and can be limited to that scope while using the same configfs interface. Rob
On Thu, Oct 07, 2021 at 03:03:43PM -0500, Rob Herring wrote: > On Thu, Oct 7, 2021 at 10:41 AM Zev Weiss <zev@bewilderbeest.net> wrote: > > > > On Thu, Oct 07, 2021 at 03:31:39AM PDT, Greg Kroah-Hartman wrote: > > >On Thu, Oct 07, 2021 at 02:05:41AM -0700, Zev Weiss wrote: > > >> On Thu, Oct 07, 2021 at 12:04:41AM PDT, Andy Shevchenko wrote: > > >> > On Thu, Oct 7, 2021 at 3:10 AM Zev Weiss <zev@bewilderbeest.net> wrote: > > >> > > This patch series is in some ways kind of a v2 for the "Dynamic > > >> > > aspeed-smc flash chips via 'reserved' DT status" series I posted > > >> > > previously [0], but takes a fairly different approach suggested by Rob > > >> > > Herring [1] and doesn't actually touch the aspeed-smc driver or > > >> > > anything in the MTD subsystem, so I haven't marked it as such. > > >> > > > > >> > > To recap a bit of the context from that series, in OpenBMC there's a > > >> > > need for certain devices (described by device-tree nodes) to be able > > >> > > to be attached and detached at runtime (for example the SPI flash for > > >> > > the host's firmware, which is shared between the BMC and the host but > > >> > > can only be accessed by one or the other at a time). > > >> > > > >> > This seems quite dangerous. Why do you need that? > > >> > > >> Sometimes the host needs access to the flash (it's the host's firmware, > > >> after all), sometimes the BMC needs access to it (e.g. to perform an > > >> out-of-band update to the host's firmware). To achieve the latter, the > > >> flash needs to be attached to the BMC, but that requires some careful > > >> coordination with the host to arbitrate which one actually has access to it > > >> (that coordination is handled by userspace, which then tells the kernel > > >> explicitly when the flash should be attached and detached). > > >> > > >> What seems dangerous? > > >> > > >> > Why can't device tree overlays be used? > > >> > > >> I'm hoping to stay closer to mainline. The OpenBMC kernel has a documented > > >> policy strongly encouraging upstream-first development: > > >> https://github.com/openbmc/docs/blob/master/kernel-development.md > > >> > > >> I doubt Joel (the OpenBMC kernel maintainer) would be eager to start > > >> carrying the DT overlay patches; I'd likewise strongly prefer to avoid > > >> carrying them myself as additional downstream patches. Hence the attempt at > > >> getting a solution to the problem upstream. > > > > > >Then why not work to get device tree overlays to be merged properly? > > TBC, it's 'just' the general purpose userspace interface to apply > overlays that's missing. > > I did suggest what's done here as overlays are kind of an overkill for > this usecase. Much easier to write to a sysfs file than write an > overlay, compile it with dtc, and provide it to the kernel all just to > enable a device. > > Perhaps this could also be supported in the driver model directly. > Given the "what about ACPI question", that is probably what should be > done unless the answer is we don't care. I think we'd just need a flag > to create devices, but not bind automatically. Or maybe abusing > driver_override can accomplish that. The driver model already allows devices to be bound/unbound from drivers, but no, it does not allow new devices to be "created" from userspace as that is a very bus-specific thing to have happen. If this is "just" a platform device, perhaps add that logic to the platform bus code? thanks, greg k-h
On 10/7/21 3:03 PM, Rob Herring wrote: > On Thu, Oct 7, 2021 at 10:41 AM Zev Weiss <zev@bewilderbeest.net> wrote: >> >> On Thu, Oct 07, 2021 at 03:31:39AM PDT, Greg Kroah-Hartman wrote: >>> On Thu, Oct 07, 2021 at 02:05:41AM -0700, Zev Weiss wrote: >>>> On Thu, Oct 07, 2021 at 12:04:41AM PDT, Andy Shevchenko wrote: >>>>> On Thu, Oct 7, 2021 at 3:10 AM Zev Weiss <zev@bewilderbeest.net> wrote: >>>>>> This patch series is in some ways kind of a v2 for the "Dynamic >>>>>> aspeed-smc flash chips via 'reserved' DT status" series I posted >>>>>> previously [0], but takes a fairly different approach suggested by Rob >>>>>> Herring [1] and doesn't actually touch the aspeed-smc driver or >>>>>> anything in the MTD subsystem, so I haven't marked it as such. >>>>>> >>>>>> To recap a bit of the context from that series, in OpenBMC there's a >>>>>> need for certain devices (described by device-tree nodes) to be able >>>>>> to be attached and detached at runtime (for example the SPI flash for >>>>>> the host's firmware, which is shared between the BMC and the host but >>>>>> can only be accessed by one or the other at a time). >>>>> >>>>> This seems quite dangerous. Why do you need that? >>>> >>>> Sometimes the host needs access to the flash (it's the host's firmware, >>>> after all), sometimes the BMC needs access to it (e.g. to perform an >>>> out-of-band update to the host's firmware). To achieve the latter, the >>>> flash needs to be attached to the BMC, but that requires some careful >>>> coordination with the host to arbitrate which one actually has access to it >>>> (that coordination is handled by userspace, which then tells the kernel >>>> explicitly when the flash should be attached and detached). >>>> >>>> What seems dangerous? >>>> >>>>> Why can't device tree overlays be used? >>>> >>>> I'm hoping to stay closer to mainline. The OpenBMC kernel has a documented >>>> policy strongly encouraging upstream-first development: >>>> https://github.com/openbmc/docs/blob/master/kernel-development.md >>>> >>>> I doubt Joel (the OpenBMC kernel maintainer) would be eager to start >>>> carrying the DT overlay patches; I'd likewise strongly prefer to avoid >>>> carrying them myself as additional downstream patches. Hence the attempt at >>>> getting a solution to the problem upstream. >>> >>> Then why not work to get device tree overlays to be merged properly? > > TBC, it's 'just' the general purpose userspace interface to apply > overlays that's missing. A fuller view of what is missing is at: https://elinux.org/Frank%27s_Evolving_Overlay_Thoughts#issues_and_what_needs_to_be_completed_--_Not_an_exhaustive_list > > I did suggest what's done here as overlays are kind of an overkill for > this usecase. Much easier to write to a sysfs file than write an > overlay, compile it with dtc, and provide it to the kernel all just to > enable a device. > > Perhaps this could also be supported in the driver model directly. > Given the "what about ACPI question", that is probably what should be > done unless the answer is we don't care. I think we'd just need a flag > to create devices, but not bind automatically. Or maybe abusing > driver_override can accomplish that. > >>> Don't work on a half-of-a-solution when the real solution is already >>> here. >>> >> >> I had been under the impression that the overlay patches had very dim >> prospects of ever being accepted and that this might be a more tractable >> alternative, but apparently was mistaken -- I'll look into what the >> outstanding issues were with that and perhaps take a stab at addressing >> them. > > What's dim is the patches allowing any modification to any part of the > DT. Any changes to a DT need to work (i.e. have some effect). For > example, randomly changing/adding/removing properties wouldn't have > any effect because they've probably already be read and used. Yes, that is a good description. > > What I've suggested before is some sort of registration of nodes > allowed to apply child nodes and properties to. That would serve the > add-on board usecases which have been the main driver of this to date. > That also got hung up on defining interface nodes to add-on boards. > Your scope is more limited and can be limited to that scope while > using the same configfs interface. > > Rob >