Mueller, Walter F J

PDP-11/70 CPU core and SoC

If the TM11 OFFLINE function is executed the backend ti_w11 crashes with an access violation. In 211bsd operation the OFFLINE function is normally not used, that's why this went undetected for some time. But other OS do issue this function, notably RSX when an ANSI labeled tape is dismounted.

Will be fixed in the next release.

TM11 OFFLINE function crashes backend

211bsd has support for RP/RM compatible 3rd party drives build by SI and CDC. These drives send a drive type of RM05 and encode the SI/CDC drive type in the SN register. The serial number returned by the rhrp firmware unfortunately matches the pattern detected by 211bsd and makes usage of real RM05 volumes at least cumbersome. Operation with RP06 type drives is not affected.

The serial number returned by rhrp will be changed in the next release to something which prevents this confusion.

RM05 detection fails in 211bsd

A race condition in the rlink backend software can lead to an abort of the event loop when three or more attention sources run under high load. One gets

    -F- 13:43:09.870897 : eloop: crashed with exception: 
          Rw11Rdma::SetupRdma: Bad state: Rdma already active

The problem first showed up when TM11 tape support was tested under 211bsd with a 'tar cvf' command, which causes high load on DL11, RHRP and TM11. It was never observed when only console plus one disk type were active.

The issue will be fixed in the next release.

ti_rri event loop aborts under heavy load with three devices

The boot command currently fails when the CPU is running with a somewhat cryptic message

   ++ wreg (002,01037) a=0001(c0.cntl   ) d=000004  s=10010001# 
      S=00000000,00000111 FAIL: ChkStat

This can happen after an OS which doesn't have a shutdown procedure, like Unix V5, is run and a different OS is booted. Current work-around is to issue a

  ti_w11 -xxx 'cpu0 cp -stop -creset'

with -xxx depending on board (as for all boot commands).

The decoding of the boot device unit number is broken, currently always unit 0 is used regardless of the specified boot device.

Both issues will be fixed in next release

boot command: fails when cpu running, unit not decoded

the rk11 controller (ibdr_rk11) doesn't issue an interrupt after a seek or drive clear command is accepted. Only the 'seek done' interrupt is issued. This causes a "rk ? csr 177400 vector 220 didn't interrupt" message during 2.11BSD boot, no other effects seen so far.
Will be fixed in next release.

rk11 doesn't interrupt after seek or drive clear command accept

fx2load_wrapper checks whether the proper firmware is already loaded and only re-loads and re-initializes the firmware if neccessary. This mechanism worked under Ubuntu 12.04, but apparently broke under Ubuntu 14.04.

As consequence fx2load_wrapper potentially always re-loads the firmware. Even though not intended behaviour, it's save, and doesn't even cost much time.

fx2load_wrapper reloads firmware even when already loaded

The Cypress FX2 firmware (under tools/fx2/src) compiles fine with sdcc 2.9 (comes with e.g. Ubuntu 12.04) but fails to compile with sdcc 3.3 (comes with Ubuntu 14.04).

Workaround: don't recompile, use the ready-to-use ihx firmware images provided under tools/fx2/bin

Will be fixed in future release

Browse

PDP-11/70 CPU core and SoC