Re: [5.4-stable PATCH 0/7] libnvdimm: Cross-arch compatible namespace alignment
On Tue, May 26, 2020 at 01:52:21PM -0700, Dan Williams wrote: > On Tue, May 26, 2020 at 1:49 PM Jeff Moyer wrote: > > > > Dan Williams writes: > > > > >> What problems with 5.4.y and 5.6.y is this series fixing > > >> that used to work before? > > > > > > The "used to work" bug fixed by this set is the fact that the kernel > > > used to force a 128MB (memory hotplug section size) alignment padding > > > on all persistent memory namespaces to enable DAX operation. The > > > support for sub-sections (2MB) dropped forced alignment padding, but > > > unfortunately introduced a regression for the case of trying to create > > > multiple unaligned namespaces. When that bug triggers namespace > > > creation for the region is disabled, iirc, previously that lockout > > > scenario was prevented. > > > > > > Jeff, can you corroborate this? > > > > So, I don't pretend to remember the exact state of brokenness for each > > iteration. :) As far as I can recall, though, the issue you describe > > with a misaligned namespace preventing further namespace creation was > > present in all kernels up until it was finally fixed. > > Well, if it was always there, then there is nothing to fix, and I > misremembered that we went backwards. > > > > I otherwise agree, if the above never worked then this can all wait > > > for v5.7 upgrades. > > > > I can test specific kernel versions if that would help out. > > Thanks for that offer, but outside of a clear regression I don't think > this meets -stable criteria. I agree, I'll drop this series from my pending-queue. thanks, greg k-h ___ Linux-nvdimm mailing list -- linux-nvdimm@lists.01.org To unsubscribe send an email to linux-nvdimm-le...@lists.01.org
Re: [5.4-stable PATCH 0/7] libnvdimm: Cross-arch compatible namespace alignment
On Tue, May 26, 2020 at 1:49 PM Jeff Moyer wrote: > > Dan Williams writes: > > >> What problems with 5.4.y and 5.6.y is this series fixing > >> that used to work before? > > > > The "used to work" bug fixed by this set is the fact that the kernel > > used to force a 128MB (memory hotplug section size) alignment padding > > on all persistent memory namespaces to enable DAX operation. The > > support for sub-sections (2MB) dropped forced alignment padding, but > > unfortunately introduced a regression for the case of trying to create > > multiple unaligned namespaces. When that bug triggers namespace > > creation for the region is disabled, iirc, previously that lockout > > scenario was prevented. > > > > Jeff, can you corroborate this? > > So, I don't pretend to remember the exact state of brokenness for each > iteration. :) As far as I can recall, though, the issue you describe > with a misaligned namespace preventing further namespace creation was > present in all kernels up until it was finally fixed. Well, if it was always there, then there is nothing to fix, and I misremembered that we went backwards. > > I otherwise agree, if the above never worked then this can all wait > > for v5.7 upgrades. > > I can test specific kernel versions if that would help out. Thanks for that offer, but outside of a clear regression I don't think this meets -stable criteria. ___ Linux-nvdimm mailing list -- linux-nvdimm@lists.01.org To unsubscribe send an email to linux-nvdimm-le...@lists.01.org
Re: [5.4-stable PATCH 0/7] libnvdimm: Cross-arch compatible namespace alignment
Dan Williams writes: >> What problems with 5.4.y and 5.6.y is this series fixing >> that used to work before? > > The "used to work" bug fixed by this set is the fact that the kernel > used to force a 128MB (memory hotplug section size) alignment padding > on all persistent memory namespaces to enable DAX operation. The > support for sub-sections (2MB) dropped forced alignment padding, but > unfortunately introduced a regression for the case of trying to create > multiple unaligned namespaces. When that bug triggers namespace > creation for the region is disabled, iirc, previously that lockout > scenario was prevented. > > Jeff, can you corroborate this? So, I don't pretend to remember the exact state of brokenness for each iteration. :) As far as I can recall, though, the issue you describe with a misaligned namespace preventing further namespace creation was present in all kernels up until it was finally fixed. > I otherwise agree, if the above never worked then this can all wait > for v5.7 upgrades. I can test specific kernel versions if that would help out. Cheers, Jeff ___ Linux-nvdimm mailing list -- linux-nvdimm@lists.01.org To unsubscribe send an email to linux-nvdimm-le...@lists.01.org
Re: [5.4-stable PATCH 0/7] libnvdimm: Cross-arch compatible namespace alignment
On Fri, May 22, 2020 at 5:00 AM Greg KH wrote: > > On Fri, May 22, 2020 at 01:58:00PM +0200, Greg KH wrote: > > On Thu, May 21, 2020 at 04:37:43PM -0700, Dan Williams wrote: > > > Hello stable team, > > > > > > These patches have been shipping in mainline since v5.7-rc1 with no > > > reported issues. They address long standing problems in libnvdimm's > > > handling of namespace provisioning relative to alignment constraints > > > including crashes trying to even load the driver on some PowerPC > > > configurations. > > > > > > I did fold one build fix [1] into "libnvdimm/region: Introduce an 'align' > > > attribute" so as to not convey the bisection breakage to -stable. > > > > > > Please consider them for v5.4-stable. They do pass the latest > > > version of the ndctl unit tests. > > > > What about 5.6.y? Any user upgrading from 5.4-stable to 5.6-stable > > would hit a regression, right? > > > > So can we get a series backported to 5.6.y as well? I need that before > > I can take this series. Yes, should be the exact same set, but I will run the regression suite to be sure. > Also, I really don't see the "bug" that this is fixing here. If this > didn't work on PowerPC before, it can continue to just "not work" until > 5.7, right? There's a mix of "never worked" and "used to work" in this set. The PowerPC case is indeed a "never worked", but I highlighted it as it was the simplest to understand. > What problems with 5.4.y and 5.6.y is this series fixing > that used to work before? The "used to work" bug fixed by this set is the fact that the kernel used to force a 128MB (memory hotplug section size) alignment padding on all persistent memory namespaces to enable DAX operation. The support for sub-sections (2MB) dropped forced alignment padding, but unfortunately introduced a regression for the case of trying to create multiple unaligned namespaces. When that bug triggers namespace creation for the region is disabled, iirc, previously that lockout scenario was prevented. Jeff, can you corroborate this? I otherwise agree, if the above never worked then this can all wait for v5.7 upgrades. ___ Linux-nvdimm mailing list -- linux-nvdimm@lists.01.org To unsubscribe send an email to linux-nvdimm-le...@lists.01.org
Re: [5.4-stable PATCH 0/7] libnvdimm: Cross-arch compatible namespace alignment
On Fri, May 22, 2020 at 01:58:00PM +0200, Greg KH wrote: > On Thu, May 21, 2020 at 04:37:43PM -0700, Dan Williams wrote: > > Hello stable team, > > > > These patches have been shipping in mainline since v5.7-rc1 with no > > reported issues. They address long standing problems in libnvdimm's > > handling of namespace provisioning relative to alignment constraints > > including crashes trying to even load the driver on some PowerPC > > configurations. > > > > I did fold one build fix [1] into "libnvdimm/region: Introduce an 'align' > > attribute" so as to not convey the bisection breakage to -stable. > > > > Please consider them for v5.4-stable. They do pass the latest > > version of the ndctl unit tests. > > What about 5.6.y? Any user upgrading from 5.4-stable to 5.6-stable > would hit a regression, right? > > So can we get a series backported to 5.6.y as well? I need that before > I can take this series. Also, I really don't see the "bug" that this is fixing here. If this didn't work on PowerPC before, it can continue to just "not work" until 5.7, right? What problems with 5.4.y and 5.6.y is this series fixing that used to work before? thanks, greg k-h ___ Linux-nvdimm mailing list -- linux-nvdimm@lists.01.org To unsubscribe send an email to linux-nvdimm-le...@lists.01.org
Re: [5.4-stable PATCH 0/7] libnvdimm: Cross-arch compatible namespace alignment
On Thu, May 21, 2020 at 04:37:43PM -0700, Dan Williams wrote: > Hello stable team, > > These patches have been shipping in mainline since v5.7-rc1 with no > reported issues. They address long standing problems in libnvdimm's > handling of namespace provisioning relative to alignment constraints > including crashes trying to even load the driver on some PowerPC > configurations. > > I did fold one build fix [1] into "libnvdimm/region: Introduce an 'align' > attribute" so as to not convey the bisection breakage to -stable. > > Please consider them for v5.4-stable. They do pass the latest > version of the ndctl unit tests. What about 5.6.y? Any user upgrading from 5.4-stable to 5.6-stable would hit a regression, right? So can we get a series backported to 5.6.y as well? I need that before I can take this series. thanks, greg k-h ___ Linux-nvdimm mailing list -- linux-nvdimm@lists.01.org To unsubscribe send an email to linux-nvdimm-le...@lists.01.org
