Instability in SDF output

[mithro]

It looks like we are still not sorting everything in the SDF output which is causing diffs in the output, see the patch below;

diff --git a/artix7/timings/GTP_CHANNEL_2.sdf b/artix7/timings/GTP_CHANNEL_2.sdf
index d846ae0..7ebaf18 100644
--- a/artix7/timings/GTP_CHANNEL_2.sdf
+++ b/artix7/timings/GTP_CHANNEL_2.sdf
@@ -144,16 +144,16 @@
         (CELLTYPE "GTPE2_CHANNEL_GTPE2_CHANNELGTPE2_CHANNEL_RXCHBONDI_GEN_TRUE_16")
         (INSTANCE GTPE2_CHANNEL)
         (TIMINGCHECK
-            (HOLD RX_DATA_WIDTH (posedge RXUSRCLK2) (0.159::0.183))
-            (SETUP RX_DATA_WIDTH (posedge RXUSRCLK2) (0.328::0.378))
+            (HOLD RX_DATA_WIDTH (posedge RXUSRCLK) (0.159::0.183))
+            (SETUP RX_DATA_WIDTH (posedge RXUSRCLK) (0.328::0.378))
         )
     )
     (CELL
         (CELLTYPE "GTPE2_CHANNEL_GTPE2_CHANNELGTPE2_CHANNEL_RXCHBONDI_GEN_TRUE_20")
         (INSTANCE GTPE2_CHANNEL)
         (TIMINGCHECK
-            (HOLD RX_DATA_WIDTH (posedge RXUSRCLK2) (0.159::0.183))
-            (SETUP RX_DATA_WIDTH (posedge RXUSRCLK2) (0.328::0.378))
+            (HOLD RX_DATA_WIDTH (posedge RXUSRCLK) (0.159::0.183))
+            (SETUP RX_DATA_WIDTH (posedge RXUSRCLK) (0.328::0.378))
         )
     )
     (CELL
diff --git a/artix7/timings/GTP_CHANNEL_3.sdf b/artix7/timings/GTP_CHANNEL_3.sdf
index d846ae0..7ebaf18 100644
--- a/artix7/timings/GTP_CHANNEL_3.sdf
+++ b/artix7/timings/GTP_CHANNEL_3.sdf
@@ -144,16 +144,16 @@
         (CELLTYPE "GTPE2_CHANNEL_GTPE2_CHANNELGTPE2_CHANNEL_RXCHBONDI_GEN_TRUE_16")
         (INSTANCE GTPE2_CHANNEL)
         (TIMINGCHECK
-            (HOLD RX_DATA_WIDTH (posedge RXUSRCLK2) (0.159::0.183))
-            (SETUP RX_DATA_WIDTH (posedge RXUSRCLK2) (0.328::0.378))
+            (HOLD RX_DATA_WIDTH (posedge RXUSRCLK) (0.159::0.183))
+            (SETUP RX_DATA_WIDTH (posedge RXUSRCLK) (0.328::0.378))
         )
     )
     (CELL
         (CELLTYPE "GTPE2_CHANNEL_GTPE2_CHANNELGTPE2_CHANNEL_RXCHBONDI_GEN_TRUE_20")
         (INSTANCE GTPE2_CHANNEL)
         (TIMINGCHECK
-            (HOLD RX_DATA_WIDTH (posedge RXUSRCLK2) (0.159::0.183))
-            (SETUP RX_DATA_WIDTH (posedge RXUSRCLK2) (0.328::0.378))
+            (HOLD RX_DATA_WIDTH (posedge RXUSRCLK) (0.159::0.183))
+            (SETUP RX_DATA_WIDTH (posedge RXUSRCLK) (0.328::0.378))
         )
     )
     (CELL

Related #776 and #815.

[mithro]

@kgugala - Can you track what is going wrong here?

[kgugala]

@mithro I'm afraid this is not sorting issue, but rather a bug in the tim2json script. It seems it does not handle correctly very complicated timings definitions as we have in e.g. GTPE2_CHANNEL.

By complicated I mean a timing which includes number of properties. I think the above timing should in fact be:

(CELLTYPE "GTPE2_CHANNEL_GTPE2_CHANNELGTPE2_CHANNEL_GEN_RXUSRCLK_TRUE_RX_DATA_WIDTH_16")
         (INSTANCE GTPE2_CHANNEL)
         (TIMINGCHECK
            (HOLD RXCHBONDI  (posedge RXUSRCLK2) (0.159::0.183))
            (SETUP RXCHBONDI (posedge RXUSRCLK2) (0.328::0.378))
         )

So from RXCHBONDI input to RXUSRCLK2 clock. with GEN_RXUSRCLK property set to true and RX_DATA_WIDTH set to 16
It looks like the script confused inputs with parameters