Opened 15 months ago

Closed 14 months ago

Last modified 14 months ago

#9145 closed bug (fixed)

hp2ps produces odd negative sawtooth

Reported by: blackdog Owned by:
Priority: normal Milestone:
Component: Profiling Version: 7.8.2
Keywords: Cc:
Operating System: Linux Architecture: Unknown/Multiple
Type of failure: Incorrect result at runtime Test Case:
Blocked By: Blocking:
Related Tickets: Differential Revisions:

Description

i'm profiling a process with about 22gb of usage before the OOMkiller gets it. the heap profile for it is very odd:

Attachments (3)

caveman.hp (74.2 KB) - added by blackdog 15 months ago.
heap profile
caveman.ps (12.2 KB) - added by blackdog 15 months ago.
generated postscript
caveman.2.ps (8.5 KB) - added by rwbarton 14 months ago.

Download all attachments as: .zip

Change History (6)

Changed 15 months ago by blackdog

heap profile

Changed 15 months ago by blackdog

generated postscript

comment:1 Changed 14 months ago by Reid Barton <rwbarton@…>

In b735883016b946372cb44b6c5d86dc36c126a8cf/ghc:

Avoid integer overflow in hp2ps (#9145)

This is slightly hackish, but hp2ps is already convoluted enough
that I don't feel bad about it.

Changed 14 months ago by rwbarton

comment:2 Changed 14 months ago by rwbarton

  • Resolution set to fixed
  • Status changed from new to closed

Here is the corrected heap profile of your RAM-guzzling program!

comment:3 Changed 14 months ago by blackdog

Heh, thank you. I've since converted it to an iteratee-based approach and it's steady on 70mb, but it's good to know that hp2ps is now safe for terrible programs everywhere :)

Note: See TracTickets for help on using tickets.