Message ID | 165063937837.27138.6911229584057659609.stgit@palantir17.mph.net (mailing list archive) |
---|---|
Headers | show |
Series | : Move Siena into a separate subdirectory | expand |
On Fri, 22 Apr 2022 15:57:43 +0100 Martin Habets wrote: > From: Martin Habets <martinh@xilinx.com> > > No changes are done, those will be done with subsequent commits. This ginormous patch does not make it thru the mail systems. I'm guessing there is a (perfectly reasonable) 1MB limit somewhere. I think you can also rework the series and combine the pure rename patches. Having the renames by header file does not substantially help review. Try to stay under the 15 patch limit.
On Sat, Apr 23, 2022 at 06:50:07AM -0700, Jakub Kicinski wrote: > On Fri, 22 Apr 2022 15:57:43 +0100 Martin Habets wrote: > > From: Martin Habets <martinh@xilinx.com> > > > > No changes are done, those will be done with subsequent commits. > > This ginormous patch does not make it thru the mail systems. > I'm guessing there is a (perfectly reasonable) 1MB limit somewhere. I think the issue is with mcdi_pcol.h, which is 1.1MB of defines generated from the hardware databases. It has grown slowely over the years. I'll split up this patch and see if I can manually cut down mcdi_pcol.h. > I think you can also rework the series and combine the pure rename > patches. Having the renames by header file does not substantially > help review. Ok, will do. I do not want to make individual patches too big. > Try to stay under the 15 patch limit. I'll probably need 2 or 3 series, and it means our Siena NICs will not work after the 1st series. Thanks for your advice, Martin
On Mon, 25 Apr 2022 08:22:57 +0100 Martin Habets wrote: > > This ginormous patch does not make it thru the mail systems. > > I'm guessing there is a (perfectly reasonable) 1MB limit somewhere. > > I think the issue is with mcdi_pcol.h, which is 1.1MB of defines > generated from the hardware databases. It has grown slowely over the > years. > I'll split up this patch and see if I can manually cut down mcdi_pcol.h. FWIW the patch did finally make an appearance on the ML but the point stands. Chiseling down the auto-generated protocol definition for a EoL part seems like a very nice solution.
On 25/04/2022 08:22, Martin Habets wrote: > I'll probably need 2 or 3 series, and it means our Siena NICs will > not work after the 1st series. Maybe instead leave the sfc.ko binding to Siena until you get to the patch that creates sfc-siena.ko? That way there's no intermediate broken state. -ed
On 2022-04-25 08:22 +0100, Martin Habets wrote: > On Sat, Apr 23, 2022 at 06:50:07AM -0700, Jakub Kicinski wrote: > > On Fri, 22 Apr 2022 15:57:43 +0100 Martin Habets wrote: > > > From: Martin Habets <martinh@xilinx.com> > > > > > > No changes are done, those will be done with subsequent commits. > > > > This ginormous patch does not make it thru the mail systems. > > I'm guessing there is a (perfectly reasonable) 1MB limit somewhere. > > I think the issue is with mcdi_pcol.h, which is 1.1MB of defines > generated from the hardware databases. It has grown slowely over the > years. > I'll split up this patch and see if I can manually cut down mcdi_pcol.h. > In this case, by using `git format-patch --find-copies-harder [...]` it greatly reduces the size of patch 03 from 1.9M to 22K and makes the actual changes easier to spot as well: [...] .../net/ethernet/sfc/{ => siena}/bitfield.h | 0 drivers/net/ethernet/sfc/{ => siena}/efx.c | 0 drivers/net/ethernet/sfc/{ => siena}/efx.h | 0 .../ethernet/sfc/{ => siena}/efx_channels.c | 148 ++++++++---------- .../ethernet/sfc/{ => siena}/efx_channels.h | 0 .../net/ethernet/sfc/{ => siena}/efx_common.c | 0 .../net/ethernet/sfc/{ => siena}/efx_common.h | 0 That being said, I don't want to discourage you from doing that rework!