On Wed, Mar 12, 2025 at 8:22 PM Doug Anderson <dianders@xxxxxxxxxxxx> wrote:
Hi,
On Wed, Mar 12, 2025 at 5:00 PM Anusha Srivatsa <asrivats@xxxxxxxxxx> wrote:
>
>
>
> On Wed, Mar 12, 2025 at 11:48 AM Doug Anderson <dianders@xxxxxxxxxxxx> wrote:
>>
>> Hi,
>>
>> On Wed, Mar 12, 2025 at 8:06 AM Anusha Srivatsa <asrivats@xxxxxxxxxx> wrote:
>> >
>> >> > @@ -106,53 +107,46 @@ static int r63353_panel_power_off(struct r63353_panel *rpanel)
>> >> > static int r63353_panel_activate(struct r63353_panel *rpanel)
>> >> > {
>> >> > struct mipi_dsi_device *dsi = rpanel->dsi;
>> >> > - struct device *dev = &dsi->dev;
>> >> > - int i, ret;
>> >> > + struct mipi_dsi_multi_context dsi_ctx = { .dsi = dsi };
>> >> > + int i;
>> >> >
>> >> > - ret = mipi_dsi_dcs_soft_reset(dsi);
>> >> > - if (ret < 0) {
>> >> > - dev_err(dev, "Failed to do Software Reset (%d)\n", ret);
>> >> > + mipi_dsi_dcs_soft_reset_multi(&dsi_ctx);
>> >> > + if (dsi_ctx.accum_err)
>> >> > goto fail;
>> >> > - }
>> >>
>> >> This changes was definitely not what the script is doing.
>> >
>> >
>> > It isnt. Using coccinelle for the major part of pattern matching and replacing the newer _multi variant API. Some handling (including a newline that it introduces) and the returns depend on a case by case basis, which had to be done manually.
>>
>> ...and now you're getting to see why I didn't think a coccinelle
>> script could fully handle this task. ;-) IMO instead of trying to get
>> a coccinelle script to do the full conversion, the right approach
>> would be to use a coccinelle script (or equivalent) to get the basics
>> done (just so you don't make any typos) and then cleanup the result
>> manually. Spending more time on the coccinelle script than it would
>> take to do the conversion manually is probably not the right approach.
>>
>> If your patch wasn't fully generated by a coccinelle script you should
>> document that in the commit message. Something like "Initial patch was
>> generated by a coccinelle script and the result was cleaned up
>> manually." If the script is too long to fit in the commit message,
>> it's fine to put it somewhere online and provide a link. "Somewhere
>> online" could easily be a mailing list post.
>>
>
> You know I have been thinking if it makes sense to have this script merged to coccinelle project and add those details in the commit log.... Was having an offline discussion with @Maxime Ripard today and he rightly pointed out that since it is too specific, once all the conversions are done , the script would not be further useable....
> About having the script online and providing a link, something link a github link?
That feels overkill. It seems like you could just send an email to
LKML or dri-devel and then link it that way. It could be in reply to
this thread or a totally separate email. lore.kernel.org links are
great.
I have been going back and forth on this for quite some time :(I have the latest v3 on a branch with the changes and I used coccinelle for most of it. I had to manually fix an extra newline that it introduced and manually handle the reset before returning, rest all was script. That makes me want to add it in the commit log unless it won't be accepted because it's too long....
Anusha
Anusha
-Doug
