Re: [PATCH RESEND v3 0/8] rtw88: improve TX performance in field

On Fri, 2021-01-15 at 09:52 +0200, Kalle Valo wrote:

Pkshih [off-list ref] writes:

quoted

To avoid frequently submitting patches results from exceeding patch size

limit.

quoted

I'd like to know the patch size limit accepted by patchwork.
As I know, the limit is about 512k, so it is expected that below patches
don't appear in patchwork
1. patch 5/5 of v1 (978K)
2. patch 6/7 of v2 (532K)

But, I don't know why the table file (patch 16/18) of rtw89 whose size is
772k can appear in patchwork. Does patchwork have different limits of
existing and new file? Moreover, if new file exceeds the limit someday,
how can I deal with it? Can I split the new file into two or more patches?

I suspect the mailing list limits the size, not patchwork. I did
directly get "[PATCH 5/5] rtw88: 8822c: update phy parameter tables to
v60" (Message-ID 20210113092312.13809-6-pkshih@realtek.com) as you added
me to CC. But I don't see it in lore, which points that linux-wireless
dropped it.

Normally these huge patches would not be applied being to big, but
updating parameter tables is a good exception to the rule and I can
commit those manually directly from my INBOX. So for huge patches I
recommend:

* move the patch as the last patch in the patchset

* the huge patch should only have changes to parameter variables, ie.
  refactor changes to the actual code to a separate patch

* add kvalo@codeaurora.org to CC

* add a big warning to the cover letter (or reply afterwards) so that I
  notice the huge patch is missing from patchwork

Would this work?

Yes, it works. Many thanks.

I would like to know if it is accepted to split the big one into two or
more patches, like my v3? Or, I should recall v1 style when I submit v4?

--
Ping-Ke