bugPSPP - Bugs: bug #58968, regression tests 403,493,545 fail...

 
 

bug #58968: regression tests 403,493,545 fail for 1.4.0 on some architectures

Submitter:  Friedrich Beckmann <beckmanf>
Submitted:  Sun 16 Aug 2020 08:57:20 PM UTC
   
 
Category:  Compilation/Portability Severity:  5 - Average
Status:  Fixed Assigned to:  None
Open/Closed:  Closed Release:  None
Effort:  0.00
* Mandatory Fields

Add a New Comment Rich Markup
   

Jump to the original submission

Mon 01 Feb 2021 06:19:26 PM UTC, comment #10: 

Ah, you're right. That output includes PSPP and its dependencies.

Since nothing at the bottom is from PSPP, there appear to be no warnings for PSPP itself.

Jeremy Lavergne <snc>
Thu 28 Jan 2021 04:28:56 PM UTC, comment #9: 

@Jeremy, are you sure this is a build log for PSPP? It seems to be all C++ files. PSPP doesn't have any C++ in it.

Ben Pfaff <blp>
Group administrator
Mon 25 Jan 2021 06:29:59 PM UTC, comment #8: 

Jumping on the arm architecture bandwagon, providing a snapshot of warning-only output from macports on arm:

https://build.macports.org/builders/ports-11_arm64-builder/builds/10014/steps/install-port/logs/warnings%20%281227%29

Jeremy Lavergne <snc>
Tue 18 Aug 2020 12:59:39 PM UTC, comment #7: 

With your last patch for regression failure 545

https://git.savannah.gnu.org/cgit/pspp.git/commit/?id=f5595c1d81afd629225d9e89fce878b25e84bc12

i made the 1.4.0-3 debian release and it seems to work! See:

https://buildd.debian.org/status/package.php?p=pspp





Friedrich Beckmann <beckmanf>
Group Member
Tue 18 Aug 2020 10:07:58 AM UTC, comment #6: 

I pushed a fix for test 545.

John Darrington <jmd>
Group administrator
Tue 18 Aug 2020 05:06:36 AM UTC, comment #5: 

After uploading 1.4.0-2 to the debian build queue we have now
two failing regression tests left:

s390x, hppa, powerpc, ppc64, sparc64: 545
545: GRAPH barchart percentage sub-categorical       FAILED (graph.at:386)
riscv64:
534: GRAPH Scatter and Histogram                     FAILED (graph.at:95) (Segfault)

Failing test 545 "GRAPH barchart percentage sub-categorical ":


## ---------------------- ##
## Detailed failed tests. ##
## ---------------------- ##

#                             -*- compilation -*-
545. graph.at:354: testing GRAPH barchart percentage sub-categorical ...
./graph.at:386: pspp --testing barchart.sps
--- -        2020-08-17 21:24:50.012127649 +0000
+++ /<<PKGBUILDDIR>>/tests/testsuite.dir/at-groups/545/stdout        2020-08-17 21:24:50.006610242 +0000
@@ -8,10 +8,10 @@
 Categorical Variables:
   Var: "penalty_favopp_x"
 Categories:
-  0 "    1.00"
-  2 "    4.00"
   3 "    5.00"
+  0 "    1.00"
   1 "    2.00"
+  2 "    4.00"
 All Categories:
 Count: 53.1634; Cat: "    1.00"
 Count: 18.7756; Cat: "    2.00"
@@ -29,10 +29,10 @@
   Var: "penalty_favopp_x"
   Var: "XYdem_racethW8"
 Categories:
-  0 "    1.00"
-  2 "    4.00"
   3 "    5.00"
+  0 "    1.00"
   1 "    2.00"
+  2 "    4.00"
 Sub-categories:
   0 "     .00"
   1 "    1.00"
545. graph.at:354: 545. GRAPH barchart percentage sub-categorical (graph.at:354): FAILED (graph.at:386)


seems to fail due to a different sequence on different architectures for HMAP_FOR_EACH in

https://git.savannah.gnu.org/cgit/pspp.git/tree/src/output/charts/barchart.c#n100




Friedrich Beckmann <beckmanf>
Group Member
Mon 17 Aug 2020 08:05:38 PM UTC, comment #4: 

The actual crash happened in memcpy due to unaligned access

So the problem started here:

https://git.savannah.gnu.org/cgit/pspp.git/tree/src/output/charts/barchart.c?id=738cf48684a7dfd5551b113b8245ec8c5ba11c35#n299

with the call of freq_clone with n_vars=3 and ended then here:

https://git.savannah.gnu.org/cgit/pspp.git/tree/src/language/stats/freq.c?id=738cf48684a7dfd5551b113b8245ec8c5ba11c35#n40

where widths[2] was accessed which ends up with segfault on i386. Maybe on 64Bit this was just luck because the values are not used in the barchart?




Friedrich Beckmann <beckmanf>
Group Member
Mon 17 Aug 2020 07:52:37 PM UTC, comment #3: 

When I look at commit 738cf48684a7dfd5551b113b8245ec8c5ba11c35  I can't understand why the problem only manifested itself on i386 and not on all achtitectures.

John Darrington <jmd>
Group administrator
Mon 17 Aug 2020 07:28:27 PM UTC, comment #2: 
Friedrich Beckmann <beckmanf>
Group Member
Sun 16 Aug 2020 09:12:05 PM UTC, comment #1: 

I did an upload of the release 1.4.0 to debian. The regression fails on some architectures:

https://buildd.debian.org/status/package.php?p=pspp

The following tests fail:

armel, armhf, i386, mipsel, hurd-i386, x32: 403, 493
s390x, ppc64, sparc64: 545
hppa, powerpc: 403,493,545

403: lexer crash due to overflow
493: CROSSTABS barchart
545: GRAPH barchart percentage sub-categorical

Friedrich Beckmann <beckmanf>
Group Member
Sun 16 Aug 2020 08:57:20 PM UTC, original submission:  

I did an upload of the release 1.4.0 to debian.

armel: 403, 493

Friedrich Beckmann <beckmanf>
Group Member

 

(Note: upload size limit is set to 16384 kB, after insertion of the required escape characters.)

Attach Files:
   
   
Comment:
   

No files currently attached

 

Depends on the following items: None found

Items that depend on this one: None found

 

Carbon-Copy List
  • -email is unavailable- added by blp (Posted a comment)
  • -email is unavailable- added by snc (Posted a comment)
  • -email is unavailable- added by jmd (Posted a comment)
  • -email is unavailable- added by beckmanf (Submitted the item)
  •  

    There are 0 votes so far. Votes easily highlight which items people would like to see resolved in priority, independently of the priority of the item set by tracker managers.

    Only logged-in users can vote.

     

    Follow 3 latest changes.

    Date Changed by Updated Field Previous Value => Replaced by
    2020-08-18 beckmanf StatusNone Fixed
        Open/ClosedOpen Closed
    2020-08-16 beckmanf Summaryregression tests 403,493, fail for 1.4.0 on some architectures - 403, 493, regression tests 403,493,545 fail for 1.4.0 on some architectures

    Back to the top

    Powered by Savane 3.13-f8d8.
    Corresponding source code