Re: [PATCH v2 03/11] lpfc: Deprecate lpfc_soft_wwn parameter
On 12/29/2016 3:31 PM, Sebastian Herbszt wrote: "not portable" might not be the term I actually meant. Let's call it server vendor dependent. Can you please elaborate on the oem platforms and side-band management you mean? I know of a solution called "ServerView Virtual-IO Manager" by Fujitsu which might fit. AFAIK it is a piece of software which communicates with the BMC and allows to change the MAC address and WWN. Of course it only works with Fujitsu servers and supported cards. I think I heard that for NICs it just changes the MAC address in the eeprom. This is not done "online" but when the server is powered on. Even though this might be a comparable case (external MAC address configuration) I think the ability to change the MAC address by the means of "ip link set dev ethX address" was not removed. Unfortunately, I can't elaborate on the platforms. Yes, there are things similar to what you describe above. Also realize that persisting the value across different cold/hot/flr resets affects boot and dump. How do you choose the desired WWN between the possible options (factory set, oem platform set, FA-WWN)? And wouldn't soft wwn always win because it is applied at last? However it might compete with FA-WWN. I wasn't yet able to find details on how it works. Sebastian Unfortunately, I can't comment on that either. It's not always last wins and that's one of the issues. I've submitted a patch to reinstated the feature - it works as well as it did before. -- james -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH v2 03/11] lpfc: Deprecate lpfc_soft_wwn parameter
James, thank you for taking the time to answer me. James Smart wrote: > Sebastian, > > "not portable" isn't the right way to describe it. It's not a > chip-architecture issue, but rather that some oem platforms have > side-band management that overrides anything that could have been > done in the os, and in ways that may not be easy to communicate back > to the driver. "not portable" might not be the term I actually meant. Let's call it server vendor dependent. Can you please elaborate on the oem platforms and side-band management you mean? I know of a solution called "ServerView Virtual-IO Manager" by Fujitsu which might fit. AFAIK it is a piece of software which communicates with the BMC and allows to change the MAC address and WWN. Of course it only works with Fujitsu servers and supported cards. I think I heard that for NICs it just changes the MAC address in the eeprom. This is not done "online" but when the server is powered on. Even though this might be a comparable case (external MAC address configuration) I think the ability to change the MAC address by the means of "ip link set dev ethX address" was not removed. > FAWN - true Brocade only - but that is the 70+% > market share of the FC switches. The current implementation already > doesn't work many situations - so there's no clear winner for having > it work everywhere. Given the inability for it always to work, the > decision was to deprecate the attribute. > > -- james How do you choose the desired WWN between the possible options (factory set, oem platform set, FA-WWN)? And wouldn't soft wwn always win because it is applied at last? However it might compete with FA-WWN. I wasn't yet able to find details on how it works. Sebastian -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH v2 03/11] lpfc: Deprecate lpfc_soft_wwn parameter
ok.. I'll submit a patch to re-add the parameters, and add an appropriate "deprecation" warning -- james On 12/28/2016 11:41 PM, Christoph Hellwig wrote: Hi James, in Linux we have a pretty clear policy to avoid breaking existing real life userspace. Given that Sebastian (and probably other users as well) make use of the soft WWN feature we can't just remove it. Which btw is what this patch does - deprecating is telling people to not use if, with the hopes of removing it in the future. I think it would be fine to deprecate it with your above rational and print a warning to the kernel log, but we can't simply remove it. -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH v2 03/11] lpfc: Deprecate lpfc_soft_wwn parameter
Hi James, in Linux we have a pretty clear policy to avoid breaking existing real life userspace. Given that Sebastian (and probably other users as well) make use of the soft WWN feature we can't just remove it. Which btw is what this patch does - deprecating is telling people to not use if, with the hopes of removing it in the future. I think it would be fine to deprecate it with your above rational and print a warning to the kernel log, but we can't simply remove it. -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH v2 03/11] lpfc: Deprecate lpfc_soft_wwn parameter
Sebastian, "not portable" isn't the right way to describe it. It's not a chip-architecture issue, but rather that some oem platforms have side-band management that overrides anything that could have been done in the os, and in ways that may not be easy to communicate back to the driver. FAWN - true Brocade only - but that is the 70+% market share of the FC switches. The current implementation already doesn't work many situations - so there's no clear winner for having it work everywhere. Given the inability for it always to work, the decision was to deprecate the attribute. -- james On 12/28/2016 10:38 AM, Sebastian Herbszt wrote: James Smart wrote: I competes, without reasonable solutions to resolve which should be what at what time, with things like fabric-assigned wwn's as well as platform-oem name assignments via platform mechanisms. Too many admin planes competing. -- james FA-WWN seems to be fabric (and Brocade?) only and "platform mechanisms" are likely not portable. Soft wwn is a simple and flexible solution which works in fabric and loop configurations and is platform and server vendor agnostic. It already works with existing HBAs on x86 and should also work on PowerPC or SPARC systems. It is useful for initiators (hardware independence/mobility) and targets (transparent failover/floating wwn). Therefore I like to ask you to revert this patch and not introduce this functional regression. Sebastian -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH v2 03/11] lpfc: Deprecate lpfc_soft_wwn parameter
James Smart wrote: > I competes, without reasonable solutions to resolve which should be > what at what time, with things like fabric-assigned wwn's as well as > platform-oem name assignments via platform mechanisms. Too many > admin planes competing. > > -- james FA-WWN seems to be fabric (and Brocade?) only and "platform mechanisms" are likely not portable. Soft wwn is a simple and flexible solution which works in fabric and loop configurations and is platform and server vendor agnostic. It already works with existing HBAs on x86 and should also work on PowerPC or SPARC systems. It is useful for initiators (hardware independence/mobility) and targets (transparent failover/floating wwn). Therefore I like to ask you to revert this patch and not introduce this functional regression. Sebastian -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH v2 03/11] lpfc: Deprecate lpfc_soft_wwn parameter
I competes, without reasonable solutions to resolve which should be what at what time, with things like fabric-assigned wwn's as well as platform-oem name assignments via platform mechanisms. Too many admin planes competing. -- james On 12/22/2016 2:00 PM, Sebastian Herbszt wrote: James Smart wrote: Deprecate lpfc_soft_wwn parameter. No longer allow override of hw-assigned wwns Signed-off-by: Dick KennedySigned-off-by: James Smart Any reason to remove this functionality? I was actually using it. Sebastian -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH v2 03/11] lpfc: Deprecate lpfc_soft_wwn parameter
James Smart wrote: > Deprecate lpfc_soft_wwn parameter. > No longer allow override of hw-assigned wwns > > Signed-off-by: Dick Kennedy> Signed-off-by: James Smart Any reason to remove this functionality? I was actually using it. Sebastian -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH v2 03/11] lpfc: Deprecate lpfc_soft_wwn parameter
On 12/20/2016 12:07 AM, James Smart wrote: > > Deprecate lpfc_soft_wwn parameter. > No longer allow override of hw-assigned wwns > > Signed-off-by: Dick Kennedy> Signed-off-by: James Smart > --- > drivers/scsi/lpfc/lpfc.h | 4 - > drivers/scsi/lpfc/lpfc_attr.c | 216 > -- > drivers/scsi/lpfc/lpfc_init.c | 16 +--- > 3 files changed, 3 insertions(+), 233 deletions(-) > Reviewed-by: Hannes Reinecke Cheers, Hannes -- Dr. Hannes ReineckeTeamlead Storage & Networking h...@suse.de +49 911 74053 688 SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg GF: F. Imendörffer, J. Smithard, J. Guild, D. Upmanyu, G. Norton HRB 21284 (AG Nürnberg) -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html