Corrupted IEEE1284 device ID string - 3.15.2

[guido]

v 3.15.2 will find my scanner. scanimage -L -> device `hpaio:/usb/HP_LaserJet_3050?serial=00CNBJ231125' is a Hewlett-Packard HP_LaserJet_3050 all-in-one. But, it will not scan. scanimage -T "hangs|polls" for a few minutes & then reports no device could be opened. My own build reports no errors, neither does hp-check. If I change the pMf var declaration in is-hp() function in hpmud.c from char * pMf; to static char * pMf;, & rebuild (make ; make install), everything works PERFECTLY! After the new pMf var declaration, I added a printf("%s\n",id); to see the IEEE1284 defined string. Here it is when functioning correctly:

MFG:Hewlett-Packard;CMD:L,PML,MLC,POSTSCRIPT,PCLXL,PCL;MDL:HP P LaserJet 3050;CLS:PRINTER;DES:Hewlett-Packa LaserJet 3050EMEM:MEM=53MB;12.4.4DL:4d,4e,1;1;CMEMENT:RES=1200x1;

And here it is without the static prefix:

MFG:Hewlett-Packard;CMD:PJL,PML,MLC,POSTSCRIPT,PCLXL,PCLDL:HP P LaserJet 3050;CLS:PRINTER;DES:Hewlett-Packa LaserJet 305050EMEM:MEM=53MB;12.4.4DL:4d,4e,1;CMEMENT:RES=1200x1;

Both strings have 173 bytes/chars & have the same beginning & end. But, in between, they are different. Note the key word MDL: in the first case is NOT in the second. Without MDL:, you cannot parse the ID string to extract the model from strstr() a few lines later.

If I add another printf("%s\n",id); near the final return statement, the entire program fails again! And I see exactly the same mangled ID string. I have also tried SANE_DEBUG_HPAIO=99 SANE_DEBUG_DLL=99 scanimage & I have used the Linux debugger gdb & have not learned anything new. Once the damaged 1284 ID string is gotten, all else fails.

I have been sucessfully building all hplip versions for about 10 years & never had any problem. My latest problem free is v 3.14.6. v 3.14.10 is like 3.15.2 despite claims made in changelogs about fixing ID strings.

Placing printf(...); in a few other places can also turn on/off the correct behavior (hpaio.c, sclpml.c, ...). The original ID string comes from the libusb_control_transfer(..) call in musb.c device_id() function at line 754; note that even the functioning ID string is corrupted. COMMENT is printed as CMEMENT + a few other errors.

I have found a resolution: enable-pp-build (even though I have no such devices) allows the scanner to work. However, the IEEE1284 ID is not exactly right, but close enough to extract a model.

Even though the latest HPLIP can be made to artificially work, there is/are bug/s that need to be fixed.

[guido (rds1944)]

I have rebuilt hplip using the just released 3.15.4 using my own AND the Slackware build scripts.

It now works!!!

(But, will it work in the next version?)