Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx> writes:
> Hi all,
>
> On Fri, 12 Jul 2024 16:43:54 +1000 Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx> wrote:
>>
>> After merging the cgroup tree, today's linux-next build (htmldocs)
>> failed like this:
>>
>> Sphinx parallel build error:
>> UnicodeDecodeError: 'utf-8' codec can't decode byte 0xfd in position 558: invalid start byte
>>
>> Caused by commit
>>
>> 704f684e15ad ("cgroup: Add Michal Koutný as a maintainer")
>>
>> I tracked this down using
>>
>> git diff stable.. | cat -v | grep -F 'M-}'
>>
>> and finding the commit that added the line that was output.
>> I tested it by building with the commit temporarily reverted. I have
>> left that commit in today's linux-next.
>>
>> Clearly (I think) there is nothing wrong with the commit, but the Sphinx
>> utf-8 decoder also clearly does not think it is valid UTF-8 :-(
>
> Actually my character map app says that it should be the two bytes 0xC3
> 0xBD in UTF-8 (it is 0x00FD in UTF-16).
As you observed, the patch is encoded in ISO-8859, not UTF8; that
doesn't fit well in a file that uses UTF8. One could argue that Sphinx
should be a bit more robust, but satisfactory results will not be had
regardless.
An encoding check might be a useful thing to have in checkpatch.pl,
methinks.
jon
