[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <f19a6d8a-c85a-963e-412e-efaa7f520453@canonical.com>
Date: Tue, 23 May 2023 09:54:06 +0200
From: Olivier Gayot <olivier.gayot@...onical.com>
To: Davidlohr Bueso <dave@...olabs.net>
Cc: Jens Axboe <axboe@...nel.dk>, linux-efi@...r.kernel.org,
linux-block@...r.kernel.org, linux-kernel@...r.kernel.org,
Daniel Bungert <daniel.bungert@...onical.com>,
Olivier Gayot <olivier.gayot@...onical.com>
Subject: [PATCH v2 0/1] block: fix conversion of GPT partition name to 7-bit
Hi Davidlohr,
This is a no-change resubmission of a patch that I originally sent to
LKML without CC-ing the relevant subsystem maintainers:
https://lore.kernel.org/all/4ae6d363-2f9b-5028-db1a-061b6f1e8fbe@canonical.com/
I apologize for the oversight in my previous submission.
--
While investigating a userspace issue, we noticed that the PARTNAME udev
property for GPT partitions is not always valid ASCII / UTF-8.
The value of the PARTNAME property for GPT partitions is initially set
by the kernel using the utf16_le_to_7bit function.
This function does a very basic conversion from UTF-16 to 7-bit ASCII by
dropping the first byte of each UTF-16 character and replacing the
remaining byte by "!" if it is not printable.
Essentially, it means that characters outside the ASCII range get
"converted" to other characters which are unrelated. Using this function
for data that is presented in userspace feels questionable and using a
proper conversion to UTF-8 would probably be preferable. However, the
patch attached does not attempt to change this design.
The patch attached actually addresses an implementation issue in the
utf16_le_to_7bit function, which causes the output of the function to
not always be valid 7-bit ASCII.
Olivier Gayot (1):
block: fix conversion of GPT partition name to 7-bit ASCII
block/partitions/efi.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
Thanks,
Olivier
Powered by blists - more mailing lists