[PR PATCH] [RFC] linux6.8: disable CONFIG_BLK_DEV_WRITE_MOUNTED

[PR PATCH] [RFC] linux6.8: disable CONFIG_BLK_DEV_WRITE_MOUNTED

[classabbyamp]

[-- Attachment #1: Type: text/plain, Size: 435 bytes --]

There is a new pull request by classabbyamp against master on the void-packages repository

https://github.com/classabbyamp/void-packages no-write-for-you
https://github.com/void-linux/void-packages/pull/50124

[RFC] linux6.8: disable CONFIG_BLK_DEV_WRITE_MOUNTED
protects against writing directly to block device nodes while they are mounted


A patch file from https://github.com/void-linux/void-packages/pull/50124.patch is attached

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: github-pr-no-write-for-you-50124.patch --]
[-- Type: text/x-diff, Size: 2026 bytes --]

From 63313ee3f34f76fb1877972a9c57d99010a44f01 Mon Sep 17 00:00:00 2001
From: classabbyamp <void@placeviolette.net>
Date: Mon, 29 Apr 2024 19:50:20 -0400
Subject: [PATCH] linux6.8: disable CONFIG_BLK_DEV_WRITE_MOUNTED

protects against writing directly to block device nodes while they are mounted
---
 srcpkgs/linux6.8/files/arm64-dotconfig  | 2 +-
 srcpkgs/linux6.8/files/i386-dotconfig   | 2 +-
 srcpkgs/linux6.8/files/x86_64-dotconfig | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/srcpkgs/linux6.8/files/arm64-dotconfig b/srcpkgs/linux6.8/files/arm64-dotconfig
index 0eea6c711dad1b..f179c61ef61c24 100644
--- a/srcpkgs/linux6.8/files/arm64-dotconfig
+++ b/srcpkgs/linux6.8/files/arm64-dotconfig
@@ -926,7 +926,7 @@ CONFIG_BLK_ICQ=y
 CONFIG_BLK_DEV_BSGLIB=y
 CONFIG_BLK_DEV_INTEGRITY=y
 CONFIG_BLK_DEV_INTEGRITY_T10=y
-CONFIG_BLK_DEV_WRITE_MOUNTED=y
+CONFIG_BLK_DEV_WRITE_MOUNTED=n
 CONFIG_BLK_DEV_ZONED=y
 CONFIG_BLK_DEV_THROTTLING=y
 # CONFIG_BLK_DEV_THROTTLING_LOW is not set
diff --git a/srcpkgs/linux6.8/files/i386-dotconfig b/srcpkgs/linux6.8/files/i386-dotconfig
index 7694b33cdd9fa7..7e6c3f359f2748 100644
--- a/srcpkgs/linux6.8/files/i386-dotconfig
+++ b/srcpkgs/linux6.8/files/i386-dotconfig
@@ -892,7 +892,7 @@ CONFIG_BLK_ICQ=y
 CONFIG_BLK_DEV_BSGLIB=y
 CONFIG_BLK_DEV_INTEGRITY=y
 CONFIG_BLK_DEV_INTEGRITY_T10=y
-CONFIG_BLK_DEV_WRITE_MOUNTED=y
+CONFIG_BLK_DEV_WRITE_MOUNTED=n
 CONFIG_BLK_DEV_ZONED=y
 CONFIG_BLK_DEV_THROTTLING=y
 # CONFIG_BLK_DEV_THROTTLING_LOW is not set
diff --git a/srcpkgs/linux6.8/files/x86_64-dotconfig b/srcpkgs/linux6.8/files/x86_64-dotconfig
index 9d35674fffd80f..de14554aae0499 100644
--- a/srcpkgs/linux6.8/files/x86_64-dotconfig
+++ b/srcpkgs/linux6.8/files/x86_64-dotconfig
@@ -969,7 +969,7 @@ CONFIG_BLK_ICQ=y
 CONFIG_BLK_DEV_BSGLIB=y
 CONFIG_BLK_DEV_INTEGRITY=y
 CONFIG_BLK_DEV_INTEGRITY_T10=y
-CONFIG_BLK_DEV_WRITE_MOUNTED=y
+CONFIG_BLK_DEV_WRITE_MOUNTED=n
 CONFIG_BLK_DEV_ZONED=y
 CONFIG_BLK_DEV_THROTTLING=y
 # CONFIG_BLK_DEV_THROTTLING_LOW is not set

Re: [RFC] linux6.8: disable CONFIG_BLK_DEV_WRITE_MOUNTED

[classabbyamp]

[-- Attachment #1: Type: text/plain, Size: 310 bytes --]

New comment by classabbyamp on void-packages repository

https://github.com/void-linux/void-packages/pull/50124#issuecomment-2083934658

Comment:
cmdline option to reënable this is `bdev_allow_write_mounted`

also, more info on the option is [here](https://www.kernelconfig.io/CONFIG_BLK_DEV_WRITE_MOUNTED)

Re: [PR PATCH] [Merged]: [RFC] linux6.8: disable CONFIG_BLK_DEV_WRITE_MOUNTED

[classabbyamp]

[-- Attachment #1: Type: text/plain, Size: 265 bytes --]

There's a merged pull request on the void-packages repository

[RFC] linux6.8: disable CONFIG_BLK_DEV_WRITE_MOUNTED
https://github.com/void-linux/void-packages/pull/50124

Description:
protects against writing directly to block device nodes while they are mounted

Re: [RFC] linux6.8: disable CONFIG_BLK_DEV_WRITE_MOUNTED

[dkwo]

[-- Attachment #1: Type: text/plain, Size: 820 bytes --]

New comment by dkwo on void-packages repository

https://github.com/void-linux/void-packages/pull/50124#issuecomment-2094118572

Comment:
With this, in linux 6.8.9 the runit boot scripts drop to an emergency shell when checking, as at this point it's still mounted ro I think:
```
=> Checking filesystems:
fsck.ext4: Device or resource busy while trying to open /dev/mapper/cryptroot-root
Filesystem mounted or opened exclusively by another program?

Cannot continue due to errors above, starting emergency shell.
```
Typing exit resumes the boot flow. The link above mentions this case:

> However there are some setups that need this capability
> like running fsck on read-only mounted root device, modifying some
> features on mounted ext4 filesystem, and similar.

What's the best way to address this?

Re: [RFC] linux6.8: disable CONFIG_BLK_DEV_WRITE_MOUNTED

[ahesford]

[-- Attachment #1: Type: text/plain, Size: 343 bytes --]

New comment by ahesford on void-packages repository

https://github.com/void-linux/void-packages/pull/50124#issuecomment-2094139054

Comment:
Does `fsck` return a different value when it fails to open the device than when it detects errors? If so, we should make the core-service script aware of that and not panic if we can't run the check. 

Re: [RFC] linux6.8: disable CONFIG_BLK_DEV_WRITE_MOUNTED

[classabbyamp]

[-- Attachment #1: Type: text/plain, Size: 260 bytes --]

New comment by classabbyamp on void-packages repository

https://github.com/void-linux/void-packages/pull/50124#issuecomment-2094205535

Comment:
we could use fsck -M

> Do not check mounted filesystems and return an exit status of 0 for mounted filesystems. 

Re: [RFC] linux6.8: disable CONFIG_BLK_DEV_WRITE_MOUNTED

[dkwo]

[-- Attachment #1: Type: text/plain, Size: 403 bytes --]

New comment by dkwo on void-packages repository

https://github.com/void-linux/void-packages/pull/50124#issuecomment-2094279822

Comment:
I think it's returining value 8.
That's an option. Alternatively, could the check be performed before mounting the filesystem?
For now, I added the option `bdev_allow_write_mounted=true` to my cmdline, as I think it's still valuable to check the root filesystem.

Re: [RFC] linux6.8: disable CONFIG_BLK_DEV_WRITE_MOUNTED

[classabbyamp]

[-- Attachment #1: Type: text/plain, Size: 347 bytes --]

New comment by classabbyamp on void-packages repository

https://github.com/void-linux/void-packages/pull/50124#issuecomment-2094350357

Comment:
8 indicates that *only* operational errors happened

and:
> The exit status returned when multiple filesystems are checked is the bit-wise OR of the exit statuses for each filesystem that is checked.