Linux system with an HP Color LaserJet Pro MFP 3302

[SteveAll]

I can confirm that this issue is still reproducible on a current Linux system with an HP Color LaserJet Pro MFP 3302.

### Environment

* Printer: **HP Color LaserJet Pro MFP 3302**
* Firmware: **6.28.1.47-202603112313**
* OS: **TUXEDO OS (Ubuntu 24.04 based)**
* KDE Plasma: **6.5.2**
* CUPS: **2.4.7**
* cups-filters: **2.0.0-0ubuntu4.1**
* libcupsfilters: **2.0.0-0ubuntu7.2**
* Connection: **IPPS (Driverless / IPP Everywhere)**

### Observed behaviour

The printer always behaves as if **two-sided-short-edge** was selected.

This happens regardless of whether the job is submitted with:

```
-o sides=two-sided-long-edge
```

or

```
-o sides=two-sided-short-edge
```

The printout is always identical and corresponds to short-edge binding.

### Verification

The printer advertises both duplex modes correctly via IPP:

```
sides-supported:
two-sided-long-edge
two-sided-short-edge
```

CUPS also submits the correct values.

For a long-edge job, the debug log contains:

```
sides=two-sided-long-edge
Duplex=DuplexNoTumble
```

For a short-edge job:

```
sides=two-sided-short-edge
Duplex=DuplexTumble
```

So CUPS and cups-filters appear to generate the correct IPP attributes.

### Additional testing

I also tested:

* Generic PostScript driver
* Generic PCL driver

Both produced incorrect duplex output (front side printed correctly, second side printed on the back of a new sheet), so they are not a usable workaround.

The old workaround of removing

```
*cupsBackSide: Rotated
```

does not apply anymore because the generated driverless PPD no longer contains this entry.

### Windows comparison

The same printer with the same firmware prints duplex correctly under Windows 11 using the official HP driver.

### Conclusion

The issue appears to be reproducible with current software.

From the tests above, it looks as if:

* the printer correctly advertises both duplex modes,
* CUPS correctly sends `two-sided-long-edge` and `two-sided-short-edge`,
* but the printer behaves as if `two-sided-short-edge` had been selected in both cases when using Driverless IPP printing.

[SteveAll]

Additional Note

PWG Raster

A separate queue was created using:

drv:///cupsfilters.drv/pwgrast.ppd

The debug log confirms that CUPS actually sends PWG Raster:

Final output format for job: image/pwg-raster
document-format="image/pwg-raster"

However, the printer only prints control characters and never produces a valid page.

Therefore PWG Raster is not a usable workaround on this printer.