diff mbox series

udev-extraconf: mount.sh: check if filesystem is supported before mounting

Message ID 20231107120017.515892-1-lukas.funke-oss@weidmueller.com
State Accepted, archived
Commit 695e0a41e3c1cb47484605934a57e7df591bd19f
Headers show
Series udev-extraconf: mount.sh: check if filesystem is supported before mounting | expand

Commit Message

Lukas Funke Nov. 7, 2023, noon UTC 
From: Lukas Funke <lukas.funke@weidmueller.com>

Check if the filesystem is supported by the kernel before trying to
mount it. Systemd-mount will mount the directories asynchronously
resulting in stale directories if the devices filesystem is not
supported.

Signed-off-by: Lukas Funke <lukas.funke@weidmueller.com>
---
 meta/recipes-core/udev/udev-extraconf/mount.sh | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

Comments

Richard Purdie Nov. 7, 2023, 12:15 p.m. UTC | #1 
On Tue, 2023-11-07 at 13:00 +0100, Lukas Funke wrote:
> From: Lukas Funke <lukas.funke@weidmueller.com>
> 
> Check if the filesystem is supported by the kernel before trying to
> mount it. Systemd-mount will mount the directories asynchronously
> resulting in stale directories if the devices filesystem is not
> supported.
> 
> Signed-off-by: Lukas Funke <lukas.funke@weidmueller.com>
> ---
>  meta/recipes-core/udev/udev-extraconf/mount.sh | 16 ++++++++++++++++
>  1 file changed, 16 insertions(+)
> 
> diff --git a/meta/recipes-core/udev/udev-extraconf/mount.sh b/meta/recipes-core/udev/udev-extraconf/mount.sh
> index 989fabc194..59563b7511 100644
> --- a/meta/recipes-core/udev/udev-extraconf/mount.sh
> +++ b/meta/recipes-core/udev/udev-extraconf/mount.sh
> @@ -36,6 +36,17 @@ do
>  	fi
>  done
>  
> +is_filesystem_supported() {
> +    list_fs=$(cat /proc/filesystems | cut -f 2 | tr '\n' ' ')
> +    for fs in ${list_fs}; do
> +        if [ "$fs" == "$ID_FS_TYPE" ]
> +        then
> +            return 0
> +        fi
> +    done
> +    return 1
> +}
> +

We once went through the init scripts and carefully minimised all the
fork() calls in them, since all that process execution overhead does
mount up and delay boot times.

I really don't know what to do, whether I should continue to care or
not. It is easy to say systems have plenty of memory/cpu and therefore,
"who cares?". Equally, systems working efficiently is nice to have...

Cheers,

Richard
Jose Quaresma Nov. 7, 2023, 12:38 p.m. UTC | #2 
Lukas Funke <lukas.funke-oss@weidmueller.com> escreveu no dia terça,
7/11/2023 à(s) 12:00:

> From: Lukas Funke <lukas.funke@weidmueller.com>
>
> Check if the filesystem is supported by the kernel before trying to
> mount it. Systemd-mount will mount the directories asynchronously
> resulting in stale directories if the devices filesystem is not
> supported.
>
> Signed-off-by: Lukas Funke <lukas.funke@weidmueller.com>
> ---
>  meta/recipes-core/udev/udev-extraconf/mount.sh | 16 ++++++++++++++++
>  1 file changed, 16 insertions(+)
>
> diff --git a/meta/recipes-core/udev/udev-extraconf/mount.sh
> b/meta/recipes-core/udev/udev-extraconf/mount.sh
> index 989fabc194..59563b7511 100644
> --- a/meta/recipes-core/udev/udev-extraconf/mount.sh
> +++ b/meta/recipes-core/udev/udev-extraconf/mount.sh
> @@ -36,6 +36,17 @@ do
>         fi
>  done
>
> +is_filesystem_supported() {
> +    list_fs=$(cat /proc/filesystems | cut -f 2 | tr '\n' ' ')
> +    for fs in ${list_fs}; do
> +        if [ "$fs" == "$ID_FS_TYPE" ]
> +        then
> +            return 0
> +        fi
> +    done
> +    return 1
> +}
> +
>  automount_systemd() {
>      name="`basename "$DEVNAME"`"
>
> @@ -64,6 +75,11 @@ automount_systemd() {
>          grep "^[[:space:]]*$tmp" /etc/fstab && return
>      done
>
> +    if ! is_filesystem_supported; then
>

I think the same can done with grep avoid string manipulations and the loop

+ if grep -q -w "$ID_FS_TYPE" /proc/filesystems; then

Jose

+        logger "mount.sh/automount" "Filesystem '$ID_FS_TYPE' on
> '${DEVNAME}' is unsupported"
> +        return
> +    fi
> +
>      [ -d "$MOUNT_BASE/$name" ] || mkdir -p "$MOUNT_BASE/$name"
>
>      MOUNT="$MOUNT -o silent"
> --
> 2.30.2
>
>
> -=-=-=-=-=-=-=-=-=-=-=-
> Links: You receive all messages sent to this group.
> View/Reply Online (#190272):
> https://lists.openembedded.org/g/openembedded-core/message/190272
> Mute This Topic: https://lists.openembedded.org/mt/102440934/5052612
> Group Owner: openembedded-core+owner@lists.openembedded.org
> Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [
> quaresma.jose@gmail.com]
> -=-=-=-=-=-=-=-=-=-=-=-
>
>
Lukas Funke Nov. 7, 2023, 12:47 p.m. UTC | #3 
Hi Richard,

On 07.11.2023 13:15, Richard Purdie wrote:
> On Tue, 2023-11-07 at 13:00 +0100, Lukas Funke wrote:
>> From: Lukas Funke <lukas.funke@weidmueller.com>
>>
>> Check if the filesystem is supported by the kernel before trying to
>> mount it. Systemd-mount will mount the directories asynchronously
>> resulting in stale directories if the devices filesystem is not
>> supported.
>>
>> Signed-off-by: Lukas Funke <lukas.funke@weidmueller.com>
>> ---
>>   meta/recipes-core/udev/udev-extraconf/mount.sh | 16 ++++++++++++++++
>>   1 file changed, 16 insertions(+)
>>
>> diff --git a/meta/recipes-core/udev/udev-extraconf/mount.sh b/meta/recipes-core/udev/udev-extraconf/mount.sh
>> index 989fabc194..59563b7511 100644
>> --- a/meta/recipes-core/udev/udev-extraconf/mount.sh
>> +++ b/meta/recipes-core/udev/udev-extraconf/mount.sh
>> @@ -36,6 +36,17 @@ do
>>   	fi
>>   done
>>   
>> +is_filesystem_supported() {
>> +    list_fs=$(cat /proc/filesystems | cut -f 2 | tr '\n' ' ')
>> +    for fs in ${list_fs}; do
>> +        if [ "$fs" == "$ID_FS_TYPE" ]
>> +        then
>> +            return 0
>> +        fi
>> +    done
>> +    return 1
>> +}
>> +
> 
> We once went through the init scripts and carefully minimised all the
> fork() calls in them, since all that process execution overhead does
> mount up and delay boot times.
> 
> I really don't know what to do, whether I should continue to care or
> not. It is easy to say systems have plenty of memory/cpu and therefore,
> "who cares?". Equally, systems working efficiently is nice to have...

Not sure if correctly interpreted, but it seems that there are some 
"frustration" vibes in your reply :)

I was not aware of the discussion but understand what you are trying to 
achieve. We can replace the 'cat' with a 'read', which should avoid the 
fork() since it's a shell-builtin (correct me if I'm wrong).

Best regards
Lukas

> 
> Cheers,
> 
> Richard
> 
>
diff mbox series

Patch

diff --git a/meta/recipes-core/udev/udev-extraconf/mount.sh b/meta/recipes-core/udev/udev-extraconf/mount.sh
index 989fabc194..59563b7511 100644
--- a/meta/recipes-core/udev/udev-extraconf/mount.sh
+++ b/meta/recipes-core/udev/udev-extraconf/mount.sh
@@ -36,6 +36,17 @@  do
 	fi
 done
 
+is_filesystem_supported() {
+    list_fs=$(cat /proc/filesystems | cut -f 2 | tr '\n' ' ')
+    for fs in ${list_fs}; do
+        if [ "$fs" == "$ID_FS_TYPE" ]
+        then
+            return 0
+        fi
+    done
+    return 1
+}
+
 automount_systemd() {
     name="`basename "$DEVNAME"`"
 
@@ -64,6 +75,11 @@  automount_systemd() {
         grep "^[[:space:]]*$tmp" /etc/fstab && return
     done
 
+    if ! is_filesystem_supported; then
+        logger "mount.sh/automount" "Filesystem '$ID_FS_TYPE' on '${DEVNAME}' is unsupported"
+        return
+    fi
+
     [ -d "$MOUNT_BASE/$name" ] || mkdir -p "$MOUNT_BASE/$name"
 
     MOUNT="$MOUNT -o silent"