Problems with Shared Drives

[Timberwoof Lupindo]

PUT file via sftp breaks shared disk on RT-11 5.5

I have a PDP11/23+ with a 256kB RAM card and a QBone for HD emulation. The Qbone passed its tests. The DPD11 checks out, too: I ran appropriate tests with XXDP; they pass. (The second DV11 test fails as expected because there's no loopback.) I can run RT-11, RSX-11, and Unix. They boot and I can poke around the filesystems. I use ethernet to talk ssh to the Qbone. I have a USB serial adapter to talk to the console. 

I start the PDP11 with 

root@qbone:~# ./rt11v5.5_dl_shared.sh -aw 22

I added a second pwr command to the end of the cmd files to kick the PDP11 when the Qbone is ready for it. This makes it ready to boot without me having to reset the PDP11 every time. 

I can sftp a simple .mac file to the Qbone and verify the PDP11 sees it. 

.dir hellor

 19-Jan-95

HELLOR.MAC     1  19-Jan-99

 1 Files, 1 Blocks

 3047 Free blocks

.type hellor.mac

        .title  Hello world program

        .mcall  .DEVICE,.EXIT,.PROTECT,.UNPROTECT,.PRINT

        .asect

[etc]

.macro hellor

caused segfault on the qbone

<<< START? DL1

D>>>./rt11v5.5_dl_shared.sh: line 3:  3382 Segmentation fault      ~/10.03_app_demo/4_deploy/demo --verbose --cmdfile rt11v5.5_dl_shared.cmd $*

The RT11 filesystem is corrupted. When I boot it hangs at .ind datime. 

I cleaned it up by deleting rt11v5.5.rl02 (which gets regenerated from the zip file).

I tried the same process and it failed differently. 

.macro hellor

caused a fault in the qbone:

[21:43:46.287543 WRN FsHost] Host: can not import DEC file , target dir  does not exist

and disabled the PDP11: 

.dir hellor

134476

@

Restart the demo program on Qbone and the PDP11 hangs as before. 

START? dl1

RT-11SJ (S) V05.05  

.ind datime

072162

@

The file is present on the shared directory. 

root@qbone:~/10.03_app_demo/5_applications/rt11.rl02/shared_rl1# ls H*.MAC

HDELAY.MAC  HELLOR.MAC HPTODK.MAC  HWDEF.MAC

I think it breaks when the assembler tries to write the object file to disk. 

What can I do to help track down the root cause? 

[Joerg Hoppe]

Hi,

yes, there's a problem with sftp.

SFTP access doesn' t trigger correctly the "file system change events" ,
which are used to detect access to the Linux directory.

See "known bugs" on https://retrocmp.com/projects/unibone/325-unibone-host-file-sharing

chatGPT "what is the problem with inotify and sftp under linux?" lists details and possible solutions ... to much stuff for now.

kind regards,

Joerg

--
You received this message because you are subscribed to the Google Groups "UniBone" group.
To unsubscribe from this group and stop receiving emails from it, send an email to unibone+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/unibone/ff077550-cfe9-4044-9c80-25b6102b23efn%40googlegroups.com.