Don't use deblob-check

Generally, this is bad, because reduces featureset of kernel.
However, we don't use any blobbed features anyway.

1. This allows much lower RAM usage.
2. Speeds up deblobbing from hours -> seconds.

This nukes blobbed files instead of replacing blobs.
This commit is contained in:
fosslinux 2021-10-12 10:43:58 +11:00 committed by Andrius Štikonas
parent f12897265c
commit 98ea0a4c2e
4 changed files with 7 additions and 90 deletions

View file

@ -1,22 +0,0 @@
SPDX-FileCopyrightText: 2021 fosslinux <fosslinux@aussies.space>
SPDX-License-Identifier: GPL-2.0-only
For some reason (old version of sed/bash?), this falsepos sed thing uses a
lot of RAM within bash and causes an overflow which is caught by bash
but prompts it to crash. As we are not using falsepos detection anyway it
is unused and hence can be set to an empty value.
--- deblob-check 2021-07-30 09:28:09.324276561 +1000
+++ deblob-check 2021-07-30 09:28:31.742449255 +1000
@@ -7232,9 +7232,7 @@
# $4 is the action for every complete input pattern.
set_sed_main () {
- falsepos=`${SED-sed} -n 's,^[+]\^*,,p' < "$regex_name" |
- ${SED-sed} -n -e 's,[$]$,\\\\([\\\\n]\\\\|$\\\\),' \
- -e '1h; 1!H; ${g;s,[\n],\\\\|,g;s,^\(..*\)$,\\\\(\1\\\\),;p;}'`
+ falsepos=
blobs=`${SED-sed} -n 's,^[-],,p' < "$regex_name" |
${SED-sed} -n -e 's,[$]$,\\\\([\\\\n]\\\\|$\\\\),' \
-e '1h; 1!H; ${g;s,[\n],\\\\|,g;s,^\(..*\)$,\\\\(\1\\\\),;p;}'`