On 2020-07-23 12:17 p.m., Kees Cook wrote:
On Wed, Jul 22, 2020 at 11:23:43PM -0700, Scott Branden wrote:
I made an adjustment in the logic of the driver with use of
request_partial_firmware_into_buf and now everything is working.
Excellent! Was there something wrong with how I ported the
request_partial_firmware_into_buf() changes? (Should the behavior be
changed in some way? I didn't change the self-tests, so I thought the
behavior matched your original series.)
The change in behaviour is fw->fw_size used to indicate the position in
the file after the partial read completes.
Your version has fw->fw_size indicate the amount of data read.
No need for you to change anything. I just need a local variable now to
keep track of where I left off in the file.
So only issue I find with this entire patch series is the problem
of security failing without the workaround below.
Yup; I'm trying to reproduce this now too...
I replied in the other email with details and the dmesg log.
_______________________________________________
kexec mailing list
kexec@xxxxxxxxxxxxxxxxxxx
http://lists.infradead.org/mailman/listinfo/kexec
