| Message ID | 1406572110-26823-3-git-send-email-mcgrof@do-not-panic.com |
|---|---|
| State | Changes Requested, archived |
| Delegated to: | David Miller |
| Headers | show |
On Mon, Jul 28, 2014 at 11:28:28AM -0700, Luis R. Rodriguez wrote: > From: "Luis R. Rodriguez" <mcgrof@suse.com> > > Tetsuo bisected and found that commit 786235ee "kthread: make > kthread_create() killable" modified kthread_create() to bail as > soon as SIGKILL is received. This is causing some issues with > some drivers and at times boot. Joseph then found that failures > occur as the systemd-udevd process sends SIGKILL to modprobe if > probe on a driver takes over 30 seconds. Because no driver should ever take that long for their probe function to return. Why not fix those drivers? > When this happens probe will fail on any driver, its why booting on > some system will fail if the driver happens to be a storage related > driver. Some folks > have suggested fixing this by modifying kthread_create() to not > leave upon SIGKILL [3], upon review Oleg rejected this change and > the discussion was punted out to systemd to see if the default > timeout could be increased from 30 seconds to 120. The opinion of > the systemd maintainers is that the driver's behavior should > be fixed [4]. Linus seems to agree [5], however more recently even > networking drivers have been reported to fail on probe since just > writing the firmware to a device and kicking it can take easy over > 60 seconds [6]. Benjamim was able to trace the issues recently > reported on cxgb4 down to the same systemd-udevd 30 second timeout [6]. Then use the async firmware interface, why is any driver taking longer than less than a second in their init function? > This is an alternative solution which enables drivers that are > known to take long to use deferred probe workqueue. This avoids > the 30 second timeout and lets us annotate drivers with long > init sequences. > > As drivers determine a component is not yet available and needs > to defer probe you'll be notified this happen upon init for each > device but now with a message such as: > > pci 0000:03:00.0: Driver cxgb4 requests probe deferral on init > > You should see one of these per struct device probed. I'm all for abusing kernel interfaces, but please, no, don't try to use the deferred init code to cover up for broken drivers. Just fix them properly, we have the interfaces to handle it properly (i.e. async firmware loading), please use it. And no PCI driver should ever need "deferred init" as the resources for such a device is already present in the system. Now if userspace is up and running yet is a different issue, one that deferred init is not there for at all, sorry. So, what drivers are having problems in their init sequence, and why aren't they using async firmware loading? thanks, greg k-h -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Mon, Jul 28, 2014 at 11:55 AM, Greg KH <gregkh@linuxfoundation.org> wrote: > So, what drivers are having problems in their init sequence, and why > aren't they using async firmware loading? Fixing drivers is one thing, fixing drivers *now* because *now* drivers are failing due to a regression is another thing and that's what we have now so lets just be clear on that. The 30 second rule is a major driver requirement change and it should not be taken slightly, all of a sudden this is a new requirement but you won't know that unless you're reading these threads or hit an issue. That's an issue in itself. The cxgb4: driver is an example where although I did propose patches to use asynch firmware loading the entire registration of the netdevice would need to be changed as well in order to get this right. In short we have to scramble now to first identify drivers that have long init sequences and then fix. There's an assumption that we can easily fix drivers, this can take time. This series, although does take advantage of a kernel interface for other uses, lets us identify these drivers on the kernel ring buffer, so we can go and address fixing them with time. Luis -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Mon, Jul 28, 2014 at 12:04 PM, Luis R. Rodriguez <mcgrof@do-not-panic.com> wrote: > On Mon, Jul 28, 2014 at 11:55 AM, Greg KH <gregkh@linuxfoundation.org> wrote: >> So, what drivers are having problems in their init sequence, and why >> aren't they using async firmware loading? > > Fixing drivers is one thing, fixing drivers *now* because *now* > drivers are failing due to a regression is another thing and that's > what we have now so lets just be clear on that. The 30 second rule is > a major driver requirement change and it should not be taken slightly, > all of a sudden this is a new requirement but you won't know that > unless you're reading these threads or hit an issue. That's an issue > in itself. > > The cxgb4: driver is an example where although I did propose patches > to use asynch firmware loading the entire registration of the > netdevice would need to be changed as well in order to get this right. > In short we have to scramble now to first identify drivers that have > long init sequences and then fix. There's an assumption that we can > easily fix drivers, this can take time. This series, although does > take advantage of a kernel interface for other uses, lets us identify > these drivers on the kernel ring buffer, so we can go and address > fixing them with time. Another thing that came up during asynch firmware review / integration on cxgb4 was that it would not be the only thing that would need to be fixed, the driver also has a ton of ports and at least as per my review the proper fix would be to use platform regiister stuff. It is a major rewrite of the driver but an example of a driver that needs quite a bit of work to meet this new 30 second driver requirement. Luis -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Mon, Jul 28, 2014 at 12:48:32PM -0700, Luis R. Rodriguez wrote: > On Mon, Jul 28, 2014 at 12:04 PM, Luis R. Rodriguez > <mcgrof@do-not-panic.com> wrote: > > On Mon, Jul 28, 2014 at 11:55 AM, Greg KH <gregkh@linuxfoundation.org> wrote: > >> So, what drivers are having problems in their init sequence, and why > >> aren't they using async firmware loading? > > > > Fixing drivers is one thing, fixing drivers *now* because *now* > > drivers are failing due to a regression is another thing and that's > > what we have now so lets just be clear on that. The 30 second rule is > > a major driver requirement change and it should not be taken slightly, > > all of a sudden this is a new requirement but you won't know that > > unless you're reading these threads or hit an issue. That's an issue > > in itself. That "regression" is something that userspace has decided to do, not anything the kernel changed, so perhaps you should just patch your modprobe and be done with it until you can fix up those drivers? And putting a horrid hack in the driver core, just because of some really bad drivers, is not ok, that's an interface _I_ will have to support for the next few decades. And it's going to take you a while to get something like this ever merged in anyway, odds are you can fix up the driver faster... > > The cxgb4: driver is an example where although I did propose patches > > to use asynch firmware loading the entire registration of the > > netdevice would need to be changed as well in order to get this right. > > In short we have to scramble now to first identify drivers that have > > long init sequences and then fix. There's an assumption that we can > > easily fix drivers, this can take time. This series, although does > > take advantage of a kernel interface for other uses, lets us identify > > these drivers on the kernel ring buffer, so we can go and address > > fixing them with time. > > Another thing that came up during asynch firmware review / integration > on cxgb4 was that it would not be the only thing that would need to be > fixed, the driver also has a ton of ports and at least as per my > review the proper fix would be to use platform regiister stuff. It is > a major rewrite of the driver but an example of a driver that needs > quite a bit of work to meet this new 30 second driver requirement. It shouldn't be using any platform driver stuff, it's a pci device, not a platform device. Why not just put the initial "register the device" in a single-shot workqueue or thread or something like that so that modprobe returns instantly back with a success and all is fine? Again, trying to hack the "deferred init" logic for PCI drivers isn't ok, I'm not going to take that into the driver core if at all possible, sorry. greg k-h -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Mon, Jul 28, 2014 at 4:46 PM, Greg KH <gregkh@linuxfoundation.org> wrote: > On Mon, Jul 28, 2014 at 12:48:32PM -0700, Luis R. Rodriguez wrote: >> On Mon, Jul 28, 2014 at 12:04 PM, Luis R. Rodriguez >> <mcgrof@do-not-panic.com> wrote: >> > On Mon, Jul 28, 2014 at 11:55 AM, Greg KH <gregkh@linuxfoundation.org> wrote: >> >> So, what drivers are having problems in their init sequence, and why >> >> aren't they using async firmware loading? >> > >> > Fixing drivers is one thing, fixing drivers *now* because *now* >> > drivers are failing due to a regression is another thing and that's >> > what we have now so lets just be clear on that. The 30 second rule is >> > a major driver requirement change and it should not be taken slightly, >> > all of a sudden this is a new requirement but you won't know that >> > unless you're reading these threads or hit an issue. That's an issue >> > in itself. > > That "regression" is something that userspace has decided to do, not > anything the kernel changed, Actually commit 786235ee seems to have been the one that caused this issue, systemd would just send the SIGKILL and that change forced a bail on probe then hence Canonical's work around to modify kthread_create() to not leave upon SIGKILL: http://thread.gmane.org/gmane.linux.ubuntu.devel.kernel.general/39123 > so perhaps you should just patch your > modprobe and be done with it until you can fix up those drivers? To ignore SIGKILL ? > And putting a horrid hack in the driver core, just because of some > really bad drivers, is not ok, that's an interface _I_ will have to > support for the next few decades. I understand, hence review. > And it's going to take you a while to get something like this ever > merged in anyway, odds are you can fix up the driver faster... That requires quite a bit of changes and commitment and again, there are quite a bit of drivers that we can run into in the community, we've just spotted 2 so far here for now. >> > The cxgb4: driver is an example where although I did propose patches >> > to use asynch firmware loading the entire registration of the >> > netdevice would need to be changed as well in order to get this right. >> > In short we have to scramble now to first identify drivers that have >> > long init sequences and then fix. There's an assumption that we can >> > easily fix drivers, this can take time. This series, although does >> > take advantage of a kernel interface for other uses, lets us identify >> > these drivers on the kernel ring buffer, so we can go and address >> > fixing them with time. >> >> Another thing that came up during asynch firmware review / integration >> on cxgb4 was that it would not be the only thing that would need to be >> fixed, the driver also has a ton of ports and at least as per my >> review the proper fix would be to use platform regiister stuff. It is >> a major rewrite of the driver but an example of a driver that needs >> quite a bit of work to meet this new 30 second driver requirement. > > It shouldn't be using any platform driver stuff, it's a pci device, not > a platform device. The general PCI stuff is already used, the reason for suggesting the platform_device_register_simple() stuff is it has tons of ports and each port will register in turn a new struct netdevice, essentially one device can end up having tons of different network devices, the platform stuff would be to allow handling each netdevice candidate separately as part of the internal driver architecture, right now its some scary loop thing that in my eyes can be very error prone. drivers/net/ethernet/8390/ne.c. This discussion: https://lkml.org/lkml/2014/6/25/815 > Why not just put the initial "register the device" in a single-shot > workqueue or thread or something like that so that modprobe returns > instantly back with a success and all is fine? That surely is possible but why not a general solution for such kludges? > Again, trying to hack the "deferred init" logic for PCI drivers isn't > ok, I'm not going to take that into the driver core if at all possible, > sorry. No need to apologize I'm looking for the best solution here after all. One userspace kludge is surely better than a one per driver while drivers are fixed for this new driver requirement. Its just kind of odd the circle of events for a kernel issue from kernel --> systemd --> modprobe as a work around. Luis -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Mon, Jul 28, 2014 at 05:26:34PM -0700, Luis R. Rodriguez wrote: > On Mon, Jul 28, 2014 at 4:46 PM, Greg KH <gregkh@linuxfoundation.org> wrote: > > On Mon, Jul 28, 2014 at 12:48:32PM -0700, Luis R. Rodriguez wrote: > >> On Mon, Jul 28, 2014 at 12:04 PM, Luis R. Rodriguez > >> <mcgrof@do-not-panic.com> wrote: > >> > On Mon, Jul 28, 2014 at 11:55 AM, Greg KH <gregkh@linuxfoundation.org> wrote: > >> >> So, what drivers are having problems in their init sequence, and why > >> >> aren't they using async firmware loading? > >> > > >> > Fixing drivers is one thing, fixing drivers *now* because *now* > >> > drivers are failing due to a regression is another thing and that's > >> > what we have now so lets just be clear on that. The 30 second rule is > >> > a major driver requirement change and it should not be taken slightly, > >> > all of a sudden this is a new requirement but you won't know that > >> > unless you're reading these threads or hit an issue. That's an issue > >> > in itself. > > > > That "regression" is something that userspace has decided to do, not > > anything the kernel changed, > > Actually commit 786235ee seems to have been the one that caused this > issue, systemd would just send the SIGKILL and that change forced a > bail on probe then hence Canonical's work around to modify > kthread_create() to not leave upon SIGKILL: > > http://thread.gmane.org/gmane.linux.ubuntu.devel.kernel.general/39123 > > > so perhaps you should just patch your > > modprobe and be done with it until you can fix up those drivers? > > To ignore SIGKILL ? Sorry, I thought this was a userspace change that caused this. As it's a kernel change, well, maybe that patch should be reverted... > > And putting a horrid hack in the driver core, just because of some > > really bad drivers, is not ok, that's an interface _I_ will have to > > support for the next few decades. > > I understand, hence review. > > > And it's going to take you a while to get something like this ever > > merged in anyway, odds are you can fix up the driver faster... > > That requires quite a bit of changes and commitment and again, there > are quite a bit of drivers that we can run into in the community, > we've just spotted 2 so far here for now. > > >> > The cxgb4: driver is an example where although I did propose patches > >> > to use asynch firmware loading the entire registration of the > >> > netdevice would need to be changed as well in order to get this right. > >> > In short we have to scramble now to first identify drivers that have > >> > long init sequences and then fix. There's an assumption that we can > >> > easily fix drivers, this can take time. This series, although does > >> > take advantage of a kernel interface for other uses, lets us identify > >> > these drivers on the kernel ring buffer, so we can go and address > >> > fixing them with time. > >> > >> Another thing that came up during asynch firmware review / integration > >> on cxgb4 was that it would not be the only thing that would need to be > >> fixed, the driver also has a ton of ports and at least as per my > >> review the proper fix would be to use platform regiister stuff. It is > >> a major rewrite of the driver but an example of a driver that needs > >> quite a bit of work to meet this new 30 second driver requirement. > > > > It shouldn't be using any platform driver stuff, it's a pci device, not > > a platform device. > > The general PCI stuff is already used, the reason for suggesting the > platform_device_register_simple() stuff is it has tons of ports and > each port will register in turn a new struct netdevice, essentially > one device can end up having tons of different network devices, the > platform stuff would be to allow handling each netdevice candidate > separately as part of the internal driver architecture, right now its > some scary loop thing that in my eyes can be very error prone. > drivers/net/ethernet/8390/ne.c. This discussion: No, don't use platform devices as a "catch-all" for when you don't want to write your own bus code. This looks like a device-specific bus, great, write the code to do that, it can be done in about 200 lines, with comments and whitespace. Only use platform devices for, wait for it, platform devices. And even then, reconsider and use something else if at all possible. > > Why not just put the initial "register the device" in a single-shot > > workqueue or thread or something like that so that modprobe returns > > instantly back with a success and all is fine? > > That surely is possible but why not a general solution for such kludges? Because the driver should be fixed. How hard would it be to do what I just suggested here, 20 lines added at most? thanks, greg k-h -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Mon, Jul 28, 2014 at 5:35 PM, Greg KH <gregkh@linuxfoundation.org> wrote: > On Mon, Jul 28, 2014 at 05:26:34PM -0700, Luis R. Rodriguez wrote: >> On Mon, Jul 28, 2014 at 4:46 PM, Greg KH <gregkh@linuxfoundation.org> wrote: >> > On Mon, Jul 28, 2014 at 12:48:32PM -0700, Luis R. Rodriguez wrote: >> >> On Mon, Jul 28, 2014 at 12:04 PM, Luis R. Rodriguez >> >> <mcgrof@do-not-panic.com> wrote: >> >> > On Mon, Jul 28, 2014 at 11:55 AM, Greg KH <gregkh@linuxfoundation.org> wrote: >> >> >> So, what drivers are having problems in their init sequence, and why >> >> >> aren't they using async firmware loading? >> >> > >> >> > Fixing drivers is one thing, fixing drivers *now* because *now* >> >> > drivers are failing due to a regression is another thing and that's >> >> > what we have now so lets just be clear on that. The 30 second rule is >> >> > a major driver requirement change and it should not be taken slightly, >> >> > all of a sudden this is a new requirement but you won't know that >> >> > unless you're reading these threads or hit an issue. That's an issue >> >> > in itself. >> > >> > That "regression" is something that userspace has decided to do, not >> > anything the kernel changed, >> >> Actually commit 786235ee seems to have been the one that caused this >> issue, systemd would just send the SIGKILL and that change forced a >> bail on probe then hence Canonical's work around to modify >> kthread_create() to not leave upon SIGKILL: >> >> http://thread.gmane.org/gmane.linux.ubuntu.devel.kernel.general/39123 >> >> > so perhaps you should just patch your >> > modprobe and be done with it until you can fix up those drivers? >> >> To ignore SIGKILL ? > > Sorry, I thought this was a userspace change that caused this. > > As it's a kernel change, well, maybe that patch should be reverted... That's certainly viable. Oleg? If its reverted we won't know which drivers are hitting over the new 30 second limit requirement imposed by userspace, which the culprit commit forces failure on probe now. This series at least would allow us to annotate which drivers are brake the 30 second init limit, and enable a work around alternative if we wish to not revert the commit. This series essentially should be considered an alternative solution to what was proposed initially by Joseph Salisbury, it may not be perfect but its a proposal. I welcome others. >> > And putting a horrid hack in the driver core, just because of some >> > really bad drivers, is not ok, that's an interface _I_ will have to >> > support for the next few decades. >> >> I understand, hence review. >> >> > And it's going to take you a while to get something like this ever >> > merged in anyway, odds are you can fix up the driver faster... >> >> That requires quite a bit of changes and commitment and again, there >> are quite a bit of drivers that we can run into in the community, >> we've just spotted 2 so far here for now. >> >> >> > The cxgb4: driver is an example where although I did propose patches >> >> > to use asynch firmware loading the entire registration of the >> >> > netdevice would need to be changed as well in order to get this right. >> >> > In short we have to scramble now to first identify drivers that have >> >> > long init sequences and then fix. There's an assumption that we can >> >> > easily fix drivers, this can take time. This series, although does >> >> > take advantage of a kernel interface for other uses, lets us identify >> >> > these drivers on the kernel ring buffer, so we can go and address >> >> > fixing them with time. >> >> >> >> Another thing that came up during asynch firmware review / integration >> >> on cxgb4 was that it would not be the only thing that would need to be >> >> fixed, the driver also has a ton of ports and at least as per my >> >> review the proper fix would be to use platform regiister stuff. It is >> >> a major rewrite of the driver but an example of a driver that needs >> >> quite a bit of work to meet this new 30 second driver requirement. >> > >> > It shouldn't be using any platform driver stuff, it's a pci device, not >> > a platform device. >> >> The general PCI stuff is already used, the reason for suggesting the >> platform_device_register_simple() stuff is it has tons of ports and >> each port will register in turn a new struct netdevice, essentially >> one device can end up having tons of different network devices, the >> platform stuff would be to allow handling each netdevice candidate >> separately as part of the internal driver architecture, right now its >> some scary loop thing that in my eyes can be very error prone. >> drivers/net/ethernet/8390/ne.c. This discussion: > > No, don't use platform devices as a "catch-all" for when you don't want > to write your own bus code. This looks like a device-specific bus, > great, write the code to do that, it can be done in about 200 lines, > with comments and whitespace. > > Only use platform devices for, wait for it, platform devices. And even > then, reconsider and use something else if at all possible. OK thanks I had asked for advice for this a while back on that old thread as I wasn't sure if that was the best. >> > Why not just put the initial "register the device" in a single-shot >> > workqueue or thread or something like that so that modprobe returns >> > instantly back with a success and all is fine? >> >> That surely is possible but why not a general solution for such kludges? > > Because the driver should be fixed. How hard would it be to do what I > just suggested here, 20 lines added at most? I appreciate the feedback, but don't look at me, I'd happy to go nutty on ripping things apart from hairy drivers, however Chelsie folks expressed concerns over major rework on the driver... so even if we did have something I expect things to take a while to bake / gain confidence from others. This also just addresses this *one* Ethernet driver, there was that SCSI driver that created the original report -- Canonical merged Joseph's fix as a work around, there is no general solution for this yet, and again with that work around you won't find which drivers run into this issue. There may be others found later so this is why I resorted to the deferred workqueue as a solution for now and to enable us to annotate which drivers need fixing as I expect getting the fix done / everyone happy can take time. Luis -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 07/29/2014 03:13 AM, Luis R. Rodriguez wrote: > On Mon, Jul 28, 2014 at 5:35 PM, Greg KH <gregkh@linuxfoundation.org> wrote: >> On Mon, Jul 28, 2014 at 05:26:34PM -0700, Luis R. Rodriguez wrote: >>> On Mon, Jul 28, 2014 at 4:46 PM, Greg KH <gregkh@linuxfoundation.org> wrote: >>>> On Mon, Jul 28, 2014 at 12:48:32PM -0700, Luis R. Rodriguez wrote: >>>>> On Mon, Jul 28, 2014 at 12:04 PM, Luis R. Rodriguez >>>>> <mcgrof@do-not-panic.com> wrote: >>>>>> On Mon, Jul 28, 2014 at 11:55 AM, Greg KH <gregkh@linuxfoundation.org> wrote: >>>>>>> So, what drivers are having problems in their init sequence, and why >>>>>>> aren't they using async firmware loading? >>>>>> >>>>>> Fixing drivers is one thing, fixing drivers *now* because *now* >>>>>> drivers are failing due to a regression is another thing and that's >>>>>> what we have now so lets just be clear on that. The 30 second rule is >>>>>> a major driver requirement change and it should not be taken slightly, >>>>>> all of a sudden this is a new requirement but you won't know that >>>>>> unless you're reading these threads or hit an issue. That's an issue >>>>>> in itself. >>>> >>>> That "regression" is something that userspace has decided to do, not >>>> anything the kernel changed, >>> >>> Actually commit 786235ee seems to have been the one that caused this >>> issue, systemd would just send the SIGKILL and that change forced a >>> bail on probe then hence Canonical's work around to modify >>> kthread_create() to not leave upon SIGKILL: >>> >>> http://thread.gmane.org/gmane.linux.ubuntu.devel.kernel.general/39123 >>> >>>> so perhaps you should just patch your >>>> modprobe and be done with it until you can fix up those drivers? >>> >>> To ignore SIGKILL ? >> >> Sorry, I thought this was a userspace change that caused this. >> >> As it's a kernel change, well, maybe that patch should be reverted... > > That's certainly viable. Oleg? > > If its reverted we won't know which drivers are hitting over the new > 30 second limit requirement imposed by userspace, which the culprit > commit forces failure on probe now. This series at least would allow > us to annotate which drivers are brake the 30 second init limit, and > enable a work around alternative if we wish to not revert the commit. > This series essentially should be considered an alternative solution > to what was proposed initially by Joseph Salisbury, it may not be > perfect but its a proposal. I welcome others. > >>>> And putting a horrid hack in the driver core, just because of some >>>> really bad drivers, is not ok, that's an interface _I_ will have to >>>> support for the next few decades. >>> >>> I understand, hence review. >>> >>>> And it's going to take you a while to get something like this ever >>>> merged in anyway, odds are you can fix up the driver faster... >>> >>> That requires quite a bit of changes and commitment and again, there >>> are quite a bit of drivers that we can run into in the community, >>> we've just spotted 2 so far here for now. >>> >>>>>> The cxgb4: driver is an example where although I did propose patches >>>>>> to use asynch firmware loading the entire registration of the >>>>>> netdevice would need to be changed as well in order to get this right. >>>>>> In short we have to scramble now to first identify drivers that have >>>>>> long init sequences and then fix. There's an assumption that we can >>>>>> easily fix drivers, this can take time. This series, although does >>>>>> take advantage of a kernel interface for other uses, lets us identify >>>>>> these drivers on the kernel ring buffer, so we can go and address >>>>>> fixing them with time. >>>>> >>>>> Another thing that came up during asynch firmware review / integration >>>>> on cxgb4 was that it would not be the only thing that would need to be >>>>> fixed, the driver also has a ton of ports and at least as per my >>>>> review the proper fix would be to use platform regiister stuff. It is >>>>> a major rewrite of the driver but an example of a driver that needs >>>>> quite a bit of work to meet this new 30 second driver requirement. >>>> >>>> It shouldn't be using any platform driver stuff, it's a pci device, not >>>> a platform device. >>> >>> The general PCI stuff is already used, the reason for suggesting the >>> platform_device_register_simple() stuff is it has tons of ports and >>> each port will register in turn a new struct netdevice, essentially >>> one device can end up having tons of different network devices, the >>> platform stuff would be to allow handling each netdevice candidate >>> separately as part of the internal driver architecture, right now its >>> some scary loop thing that in my eyes can be very error prone. >>> drivers/net/ethernet/8390/ne.c. This discussion: >> >> No, don't use platform devices as a "catch-all" for when you don't want >> to write your own bus code. This looks like a device-specific bus, >> great, write the code to do that, it can be done in about 200 lines, >> with comments and whitespace. >> >> Only use platform devices for, wait for it, platform devices. And even >> then, reconsider and use something else if at all possible. > > OK thanks I had asked for advice for this a while back on that old > thread as I wasn't sure if that was the best. > >>>> Why not just put the initial "register the device" in a single-shot >>>> workqueue or thread or something like that so that modprobe returns >>>> instantly back with a success and all is fine? >>> >>> That surely is possible but why not a general solution for such kludges? >> >> Because the driver should be fixed. How hard would it be to do what I >> just suggested here, 20 lines added at most? > > I appreciate the feedback, but don't look at me, I'd happy to go nutty > on ripping things apart from hairy drivers, however Chelsie folks > expressed concerns over major rework on the driver... so even if we > did have something I expect things to take a while to bake / gain > confidence from others. > > This also just addresses this *one* Ethernet driver, there was that > SCSI driver that created the original report -- Canonical merged > Joseph's fix as a work around, there is no general solution for this > yet, and again with that work around you won't find which drivers run > into this issue. There may be others found later so this is why I > resorted to the deferred workqueue as a solution for now and to enable > us to annotate which drivers need fixing as I expect getting the fix > done / everyone happy can take time. > Well ... from my POV there are two issues here: The one is that systemd essentially nailed its internal worker timeout to 30 seconds. And there is no way of modifying that short of recompiling. Which should be fixed, so that at least one can modify this timeout. The other one is that systemd killing a worker by sending SIGKILL, which will kill modprobe terminally. Which definitely needs to be fixed. But if we have both issues resolved (eg by allowing udevd to use a longer timeout and revert the latter patch) we can identify the offending drivers _and_ get the system to boot by simply adding a kernel commandline parameter. Which is _far_ preferrable from a maintenance perspective. Users tend to become annoyed if their system doesn't boot ... Cheers, Hannes
From: "Luis R. Rodriguez" <mcgrof@do-not-panic.com> Date: Mon, 28 Jul 2014 11:28:28 -0700 > Tetsuo bisected and found that commit 786235ee "kthread: make > kthread_create() killable" modified kthread_create() to bail as > soon as SIGKILL is received. This is causing some issues with > some drivers and at times boot. Joseph then found that failures > occur as the systemd-udevd process sends SIGKILL to modprobe if > probe on a driver takes over 30 seconds. When this happens probe > will fail on any driver, its why booting on some system will fail > if the driver happens to be a storage related driver. Some folks > have suggested fixing this by modifying kthread_create() to not > leave upon SIGKILL [3], upon review Oleg rejected this change and > the discussion was punted out to systemd to see if the default > timeout could be increased from 30 seconds to 120. The opinion of > the systemd maintainers is that the driver's behavior should > be fixed [4]. Linus seems to agree [5], however more recently even > networking drivers have been reported to fail on probe since just > writing the firmware to a device and kicking it can take easy over > 60 seconds [6]. Benjamim was able to trace the issues recently > reported on cxgb4 down to the same systemd-udevd 30 second timeout [6]. > > This is an alternative solution which enables drivers that are > known to take long to use deferred probe workqueue. This avoids > the 30 second timeout and lets us annotate drivers with long > init sequences. > > As drivers determine a component is not yet available and needs > to defer probe you'll be notified this happen upon init for each > device but now with a message such as: > > pci 0000:03:00.0: Driver cxgb4 requests probe deferral on init > > You should see one of these per struct device probed. It seems we're still discussing this. I think I understand all of the underlying issues, and what I'll say is that perhaps we should use what Greg KH requested but via a helper that is easy to grep for. I don't care if it's something like "module_long_probe_init()" and "module_long_probe_exit()", but it just needs to be some properly named interface which does the whole kthread or whatever bit. Thanks. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Jul 30, 2014 at 03:11:07PM -0700, David Miller wrote: > From: "Luis R. Rodriguez" <mcgrof@do-not-panic.com> > Date: Mon, 28 Jul 2014 11:28:28 -0700 > > > Tetsuo bisected and found that commit 786235ee "kthread: make > > kthread_create() killable" modified kthread_create() to bail as > > soon as SIGKILL is received. This is causing some issues with > > some drivers and at times boot. Joseph then found that failures > > occur as the systemd-udevd process sends SIGKILL to modprobe if > > probe on a driver takes over 30 seconds. When this happens probe > > will fail on any driver, its why booting on some system will fail > > if the driver happens to be a storage related driver. Some folks > > have suggested fixing this by modifying kthread_create() to not > > leave upon SIGKILL [3], upon review Oleg rejected this change and > > the discussion was punted out to systemd to see if the default > > timeout could be increased from 30 seconds to 120. The opinion of > > the systemd maintainers is that the driver's behavior should > > be fixed [4]. Linus seems to agree [5], however more recently even > > networking drivers have been reported to fail on probe since just > > writing the firmware to a device and kicking it can take easy over > > 60 seconds [6]. Benjamim was able to trace the issues recently > > reported on cxgb4 down to the same systemd-udevd 30 second timeout [6]. > > > > This is an alternative solution which enables drivers that are > > known to take long to use deferred probe workqueue. This avoids > > the 30 second timeout and lets us annotate drivers with long > > init sequences. > > > > As drivers determine a component is not yet available and needs > > to defer probe you'll be notified this happen upon init for each > > device but now with a message such as: > > > > pci 0000:03:00.0: Driver cxgb4 requests probe deferral on init > > > > You should see one of these per struct device probed. > > It seems we're still discussing this. > > I think I understand all of the underlying issues, and what I'll say > is that perhaps we should use what Greg KH requested but via a helper > that is easy to grep for. > > I don't care if it's something like "module_long_probe_init()" and > "module_long_probe_exit()", but it just needs to be some properly > named interface which does the whole kthread or whatever bit. I've tested the alternative kthread_run() proposal but unfortunately it does not help resolve the issue, the timeout is still hit and a SIGKILL still kills the driver probe. Please let me know how you'd all like us to proceed, these defer probe patches do help cure the issue though. I should also note that these work around patches can only be done once we already know a driver fails to go over the timeout, root causing and associating driver issues to the timeout has been very difficult with a few drivers already, for this reason I've submitted a change for systemd to issue a warning instead of killing kmod usage on udev after a timeout, that would make this regression non-fatal, and let us more easily then hunt drivers that need fixing much easily [0] [1]. As noted we'd still want to have drivers easily annotated which require fixing, this orignal series would allow us to do that by hunting for delay_probe. If there alternative and preferred strategies please let me know. [0] http://lists.freedesktop.org/archives/systemd-devel/2014-August/021812.html [1] http://lists.freedesktop.org/archives/systemd-devel/2014-August/021821.html Luis -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Sat, Aug 09, 2014 at 06:41:19PM +0200, Luis R. Rodriguez wrote: > On Wed, Jul 30, 2014 at 03:11:07PM -0700, David Miller wrote: > > From: "Luis R. Rodriguez" <mcgrof@do-not-panic.com> > > Date: Mon, 28 Jul 2014 11:28:28 -0700 > > > > > Tetsuo bisected and found that commit 786235ee "kthread: make > > > kthread_create() killable" modified kthread_create() to bail as > > > soon as SIGKILL is received. This is causing some issues with > > > some drivers and at times boot. Joseph then found that failures > > > occur as the systemd-udevd process sends SIGKILL to modprobe if > > > probe on a driver takes over 30 seconds. When this happens probe > > > will fail on any driver, its why booting on some system will fail > > > if the driver happens to be a storage related driver. Some folks > > > have suggested fixing this by modifying kthread_create() to not > > > leave upon SIGKILL [3], upon review Oleg rejected this change and > > > the discussion was punted out to systemd to see if the default > > > timeout could be increased from 30 seconds to 120. The opinion of > > > the systemd maintainers is that the driver's behavior should > > > be fixed [4]. Linus seems to agree [5], however more recently even > > > networking drivers have been reported to fail on probe since just > > > writing the firmware to a device and kicking it can take easy over > > > 60 seconds [6]. Benjamim was able to trace the issues recently > > > reported on cxgb4 down to the same systemd-udevd 30 second timeout [6]. > > > > > > This is an alternative solution which enables drivers that are > > > known to take long to use deferred probe workqueue. This avoids > > > the 30 second timeout and lets us annotate drivers with long > > > init sequences. > > > > > > As drivers determine a component is not yet available and needs > > > to defer probe you'll be notified this happen upon init for each > > > device but now with a message such as: > > > > > > pci 0000:03:00.0: Driver cxgb4 requests probe deferral on init > > > > > > You should see one of these per struct device probed. > > > > It seems we're still discussing this. > > > > I think I understand all of the underlying issues, and what I'll say > > is that perhaps we should use what Greg KH requested but via a helper > > that is easy to grep for. > > > > I don't care if it's something like "module_long_probe_init()" and > > "module_long_probe_exit()", but it just needs to be some properly > > named interface which does the whole kthread or whatever bit. > > I've tested the alternative kthread_run() proposal but unfortunately it > does not help resolve the issue, the timeout is still hit and a SIGKILL > still kills the driver probe. Please let me know how you'd all like us > to proceed, these defer probe patches do help cure the issue though. Why doesn't it work? Doesn't modprobe come right back and the init sequence still takes a while to run? What exactly fails? thanks, greg k-h -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Greg KH wrote: > Why doesn't it work? Doesn't modprobe come right back and the init > sequence still takes a while to run? What exactly fails? I guess ... > @@ -5429,9 +5429,19 @@ mptsas_init(void) > return error; > } > > +static struct task_struct *init_thread; > + > +static int __init > +mptsas_init(void) > +{ > + init_thread = kthread_run(mptsas_real_init, NULL, "mptsas_init"); > + return 0; > +} > + > static void __exit > mptsas_exit(void) > { > + kthread_stop(init_thread); > pci_unregister_driver(&mptsas_driver); > sas_release_transport(mptsas_transport_template); > > kthread_run() can fail. sas_attach_transport() and/or pci_register_driver() in mptsas_real_init() can fail. Caller process may fail to continue if sas_attach_transport() and pci_register_driver() in mptsas_real_init() has not completed yet. kthread_stop() must not be called when kthread_run() failed. pci_unregister_driver() and/or sas_release_transport() must not be called when mptsas_real_init() did not return 0 (or has not returned 0 yet). -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/base/dd.c b/drivers/base/dd.c index 9947c2e..d59e4b5 100644 --- a/drivers/base/dd.c +++ b/drivers/base/dd.c @@ -159,6 +159,9 @@ static void driver_deferred_probe_add(struct device *dev) list_add_tail(&dev->p->deferred_probe, &deferred_probe_pending_list); } mutex_unlock(&deferred_probe_mutex); + + if (driver_deferred_probe_enable) + driver_deferred_probe_trigger(); } void driver_deferred_probe_del(struct device *dev) @@ -374,6 +377,19 @@ void wait_for_device_probe(void) } EXPORT_SYMBOL_GPL(wait_for_device_probe); +static int __driver_probe_device(struct device_driver *drv, struct device *dev) +{ + if (drv->delay_probe && !dev->init_delayed_probe) { + dev_info(dev, "Driver %s requests probe deferral on init\n", + drv->name); + dev->init_delayed_probe = true; + driver_deferred_probe_add(dev); + return -EPROBE_DEFER; + } + + return really_probe(dev, drv); +} + /** * driver_probe_device - attempt to bind device & driver together * @drv: driver to bind a device to @@ -396,7 +412,7 @@ int driver_probe_device(struct device_driver *drv, struct device *dev) drv->bus->name, __func__, dev_name(dev), drv->name); pm_runtime_barrier(dev); - ret = really_probe(dev, drv); + ret = __driver_probe_device(drv, dev); pm_request_idle(dev); return ret; diff --git a/include/linux/device.h b/include/linux/device.h index af424ac..c317dfa 100644 --- a/include/linux/device.h +++ b/include/linux/device.h @@ -200,6 +200,9 @@ extern struct klist *bus_get_device_klist(struct bus_type *bus); * @owner: The module owner. * @mod_name: Used for built-in modules. * @suppress_bind_attrs: Disables bind/unbind via sysfs. + * @delay_probe: this driver is requesting a deferred probe since + * initialization. This can be desirable if its known the device probe + * or initialization takes more than 30 seconds. * @of_match_table: The open firmware table. * @acpi_match_table: The ACPI match table. * @probe: Called to query the existence of a specific device, @@ -233,6 +236,7 @@ struct device_driver { const char *mod_name; /* used for built-in modules */ bool suppress_bind_attrs; /* disables bind/unbind via sysfs */ + bool delay_probe; /* requests deferred probe */ const struct of_device_id *of_match_table; const struct acpi_device_id *acpi_match_table; @@ -715,6 +719,8 @@ struct acpi_dev_node { * * @offline_disabled: If set, the device is permanently online. * @offline: Set after successful invocation of bus type's .offline(). + * @init_delayed_probe: lets the coore keep track if the device has already + * gone through a delayed probe upon init. * * At the lowest level, every device in a Linux system is represented by an * instance of struct device. The device structure contains the information @@ -793,6 +799,7 @@ struct device { bool offline_disabled:1; bool offline:1; + bool init_delayed_probe:1; }; static inline struct device *kobj_to_dev(struct kobject *kobj)