Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Incorrect PLL IN_USE feature #1141

Closed
litghost opened this issue Nov 12, 2019 · 11 comments · Fixed by #1142
Closed

Incorrect PLL IN_USE feature #1141

litghost opened this issue Nov 12, 2019 · 11 comments · Fixed by #1142
Assignees
@mkurc-ant

Comments

@litghost
Copy link
Contributor

f4pga/f4pga-arch-defs#1150 failed as a result of a bad segbit definition:

FAILED: cd /tmpfs/src/github/symbiflow-arch-defs-presubmit-xc7/build/xc7/tests/pll && /usr/bin/cmake -E env PYTHONPATH=/tmpfs/src/github/symbiflow-arch-defs-presubmit-xc7/build/env/conda/lib/python3.7/site-packages:/tmpfs/src/github/symbiflow-arch-defs-presubmit-xc7/third_party/prjxray:/tmpfs/src/github/symbiflow-arch-defs-presubmit-xc7/third_party/prjxray/third_party/fasm /tmpfs/src/github/symbiflow-arch-defs-presubmit-xc7/build/env/conda/bin/python3 /tmpfs/src/github/symbiflow-arch-defs-presubmit-xc7/third_party/prjxray/utils/fasm2frames.py --db-root /tmpfs/src/github/symbiflow-arch-defs-presubmit-xc7/third_party/prjxray-db/artix7 --sparse --emit_pudc_b_pullup /tmpfs/src/github/symbiflow-arch-defs-presubmit-xc7/build/xc7/tests/pll/pll_ext_basys3_x1y0/artix7-xc7a50t-roi-virt-xc7a50t-test/top.fasm /tmpfs/src/github/symbiflow-arch-defs-presubmit-xc7/build/xc7/tests/pll/pll_ext_basys3_x1y0/artix7-xc7a50t-roi-virt-xc7a50t-test/top.frames
Traceback (most recent call last):
  File "/tmpfs/src/github/symbiflow-arch-defs-presubmit-xc7/third_party/prjxray/utils/fasm2frames.py", line 199, in <module>
    main()
  File "/tmpfs/src/github/symbiflow-arch-defs-presubmit-xc7/third_party/prjxray/utils/fasm2frames.py", line 195, in main
    emit_pudc_b_pullup=args.emit_pudc_b_pullup)
  File "/tmpfs/src/github/symbiflow-arch-defs-presubmit-xc7/third_party/prjxray/utils/fasm2frames.py", line 123, in run
    assembler.parse_fasm_filename(filename_in, extra_features=extra_features)
  File "/tmpfs/src/github/symbiflow-arch-defs-presubmit-xc7/third_party/prjxray/prjxray/fasm_assembler.py", line 174, in parse_fasm_filename
    self.add_fasm_line(line, missing_features)
  File "/tmpfs/src/github/symbiflow-arch-defs-presubmit-xc7/third_party/prjxray/prjxray/fasm_assembler.py", line 167, in add_fasm_line
    self.enable_feature(tile, feature, address, line_str)
  File "/tmpfs/src/github/symbiflow-arch-defs-presubmit-xc7/third_party/prjxray/prjxray/fasm_assembler.py", line 129, in enable_feature
    update_segbit(bit)
  File "/tmpfs/src/github/symbiflow-arch-defs-presubmit-xc7/third_party/prjxray/prjxray/fasm_assembler.py", line 115, in update_segbit
    self.frame_clear(frame_addr, word_addr, bit_index, line)
  File "/tmpfs/src/github/symbiflow-arch-defs-presubmit-xc7/third_party/prjxray/prjxray/fasm_assembler.py", line 96, in frame_clear
    self.frames_line[key],
prjxray.fasm_assembler.FasmInconsistentBits: FASM line "CMT_TOP_L_UPPER_T_X106Y44.CMT_TOP_R_UPPER_T_PLLE2_CLKIN2.CMT_TOP_L_UPPER_T_FREQ_BB0" wanted to clear bit (4199709, 75, 8) but was set by FASM line "CMT_TOP_L_UPPER_T_X106Y44.PLLE2.IN_USE"
ninja: build stopped: subcommand failed.
@litghost
Copy link
Contributor Author

IN_USE:

CMT_TOP_L_UPPER_T.PLLE2.IN_USE 28_05 28_10 28_16 28_42 28_46 28_184 28_187 28_190 28_234 28_248 28_251 28_252 28_560 28_590 28_591 28_592 28_595 28_596 28_736 28_753 28_754 28_755 28_756 28_758 28_759 28_760 28_761 28_763 28_764 28_765 28_766 28_767 28_771 28_774 28_775 28_779 28_780 28_783 28_784 28_785 28_786 28_787 28_789 28_790 28_791 28_792 28_794 28_795 28_796 28_797 28_800 28_803 28_806 28_809 28_813 28_815 29_08 29_16 29_45 29_46 29_47 29_236 29_249 29_250 29_251 29_531 29_535 29_557 29_589 29_590 29_593 29_594 29_595 29_753 29_754 29_756 29_758 29_759 29_761 29_765 29_766 29_768 29_777 29_779 29_780 29_781 29_782 29_783 29_784 29_785 29_787 29_791 29_792 29_793 29_794 29_795 29_797 29_798 29_799 29_800 29_803 29_804 29_806 29_807 29_808 29_809 29_810 29_813

versus

CMT_TOP_L_UPPER_T.CMT_TOP_R_UPPER_T_PLLE2_CLKIN2.CMT_TOP_L_UPPER_T_FREQ_BB0 !28_08 !29_07 !29_08

@litghost
Copy link
Contributor Author

litghost commented Nov 12, 2019 •
edited
Loading

Looks like 29_08 is in IN_USE, but cleared by the FREQ_BB0->CLKIN2.

@mkurc-ant
Copy link
Contributor

That's strange. I've runned both 032 and 034 fuzzers locally (current master 65ccddb) and none of them found the bit 29_08 in the IN_USE feature. ('034' solves the IN_USE as a byproduct).

Investigating...

@mkurc-ant
Copy link
Contributor

The solution for IN_USE you posted contains additional bits w.r.t. the solution I've got: 28_10 and 29_08.

Please provide SHA of the prjxray-db you are using and if possible SHA of the prjxray repo used to generate it.

@acomodi
Copy link
Contributor

acomodi commented Nov 13, 2019

Even though this does not regard the bits collision issue, I have seen that, by driving the PLL directly from the IBUF_clk net lets the CMT_TOP_L_UPPER_T_X106Y44.CMT_TOP_R_UPPER_T_PLLE2_CLKIN2.CMT_TOP_L_UPPER_T_FREQ_BB0 pip to be used.

Instead, if the CLKIN comes from a BUFG, VPR chooses a different path, avoiding the generation of this failure.

@mkurc-ant
Copy link
Contributor

@acomodi By "IBUF" you mean a regular input buffer, not the dedicated clock input buffer? Because we do not have bits for clock input buffers that are in HCLK_IOI3 tile.

Could you elaborate what is the failure?

@acomodi
Copy link
Contributor

acomodi commented Nov 13, 2019

The failure is the one related to this issue: if the CMT_TOP_L_UPPER_T_X106Y44.CMT_TOP_R_UPPER_T_PLLE2_CLKIN2.CMT_TOP_L_UPPER_T_FREQ_BB0 pip is chosen to route to one of the clock inputs of the PLL, than there is a conflict as the bits for this pip are cleared and some of the same bits are enabled by the PLLE_ADV.IN_USE feature.

As regards the IBUF clk net, i mean the dedicated clock input buffer (the W5 IOPAD).

@litghost
Copy link
Contributor Author

litghost commented Nov 13, 2019 •
edited
Loading

Please provide SHA of the prjxray-db you are using and if possible SHA of the prjxray repo used to generate it.

f4pga/prjxray-db@ca781e7 : https://github.com/SymbiFlow/prjxray-db/blob/master/artix7/segbits_cmt_top_l_upper_t.db#L74
aa9393b

litghost added a commit to litghost/symbiflow-arch-defs that referenced this issue Nov 13, 2019
@litghost
Disable pll tests until f4pga/prjxray#1141 is resolved.
3ba16d6 
Signed-off-by: Keith Rothman <537074+litghost@users.noreply.github.com>
@litghost
Copy link
Contributor Author

prjxray master is still showing 29_08 in the IN_USE.

@mkurc-ant
Copy link
Contributor

Ok, I'm going to run a full build of the db locally (clean build, different machine) and see where the problem is.

@mkurc-ant mkurc-ant mentioned this issue Nov 14, 2019
Merged
@mkurc-ant
Copy link
Contributor

@litghost #1142 I filtered out conflicting bits in 032.

When I ran the "fasm diff" target for the PLL, the bitstream was decoded to fasm without any error but then Vivado complained that some clock routes are "illegal" and in order to be there require CLOCK_DEDICATED_ROUTE = false.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@mkurc-ant mkurc-ant

Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

Update to 032 which fixes conflict with 034 antmicro/prjxray
3 participants
@litghost @acomodi @mkurc-ant