mount using SMB1 only despite vers=2.1 specified

[ColinTaylor]

(it all started with the BACKUPMON tbh)

Just write you own backup script. Mine is 19 lines long and uses lftp to send the tar files to my server.

 

[Jeffrey Young]

Just write you own backup script. Mine is 19 lines long and uses lftp to send the tar files to my server.

Agreed. That BACKUPMON was built around my script, which is also about 20 lines in total. Another option would be to use rsync.

 

[Phantomski]

Just write you own backup script. Mine is 19 lines long and uses lftp to send the tar files to my server.

That sounds like what I'll have to do anyway.

I was trying to investigate this a little deeper - I just couldn't understand why the mount was an old limited version. New one can be installed from entware, but it also needs cifs-utils, that provides mount.cifs module and that can't be installed from entware. I have tried to compile it from the source (which is available at entware repo), but while I can compile other packages, cifs-utils just wouldn't budge. I don't know the reasons behind this, but it seems abandoned in both entware and OpenWrt, so I have reached the dead end for now.

Update: after I've managed to compile mount.cifs I have realised @ColinTaylor was right all along, of course It is the kernel module cifs.ko that is the problem. Now I am well out of my depth here (sorry), but I have noticed a few differences in some of Merlin's releases for this module.
Check src-rt-6.x.4708 vs src-rt-5.02hnd, notably the last missing bit:

cifs-$(CONFIG_CIFS_SMB2) += smb2ops.o smb2maperror.o smb2transport.o \
                            smb2misc.o smb2pdu.o smb2inode.o smb2file.o

Is that the regression that happened? It is perhaps reversible @RMerlin in the next release, or is the problem bigger in terms of dependencies or the CPU capability, etc.?

 

[RMerlin]

Is that the regression that happened? It is perhaps reversible @RMerlin in the next release, or is the problem bigger in terms of dependencies or the CPU capability, etc.?

You are comparing kernel 2.6.36 against kernel 4.1. Very different capabilities.

 

[Phantomski]

You are comparing kernel 2.6.36 against kernel 4.1. Very different capabilities.

You're right of course. Just went through the kernel history and it seems only 3.7 properly introduced SMB2
Just before I completely abandon the idea of SMB2+ mount.cifs in our 2.6.36 arm7l kernel on AC88U, is there any future hope of backporting SMB2 support? Anyone attempted that in other projects?

This was all just out of interest and little experimentation, I'm planning to upgrade to AX88U Pro soon-ish, which is 4.1.x if I'm right, which should sort this particular "problem". Thanks for quick heads up.

 

[JGrana]

I'm running an AX88U Pro and as long as you add:

/sbin/modprobe md4

to /jffs/scripts/services-start, you will be good to go!

 

[RMerlin]

is there any future hope of backporting SMB2 support?

None. For starter, most of these 2.6.36 models are now EOL. So even if it was technically possible, it's not worth the effort.

 

[ColinTaylor]

I'm running an AX88U Pro and as long as you add:

/sbin/modprobe md4

to /jffs/scripts/services-start, you will be good to go!

This is not applicable to the router being discussed (and was also mentioned in post #1). See the previous posts in this thread.

 

[JGrana]

This is not applicable to the router being discussed (and was also mentioned in post #1). See the previous posts in this thread.

Sorry, I should have been more clear. I believe he will have to do this when he gets his new AX88U Pro router.
I still am doing it with mine…