BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Diagnosing Production-Run Concurrency-Bug Failures
DTSTART:20140203T103000
DTEND:20140203T113000
DTSTAMP:20260407T102529Z
UID:db58dfe12912198b47837b935f589dab808eff9ca0000bd13b7b114a
CATEGORIES:Conferences - Seminars
DESCRIPTION:Shan LU\nFailures caused by software bugs are widespread in pr
 oduction runs\, causing severe losses for end users. Unfortunately\, diagn
 osing production-run failures\, especially failures caused by concurrency 
 bugs in multi-threaded software\, is challenging. Existing work cannot sat
 isfy privacy\, run-time overhead\, diagnosis capability\, and diagnosis la
 tency requirements all at once.\nThis talk will present a series of attemp
 ts from our group to address the above challenges. Our first attempt\, cal
 led CCI\, applies the cooperative bug isolation (CBI) approach\, which was
  initially designed for sequential bugs\, to concurrency bugs. Our careful
 ly designed interleaving predicates and sampling schemes allow CCI to diag
 nose a wide variety of concurrency-bug failures with decent overhead. Our 
 second attempt\, called PBI\, further improves the performance and preserv
 es the diagnosis capability of CCI through a novel use of hardware perform
 ance counters. Our final attempt\, called LXR\, addresses the long diagnos
 is latency problem of CCI and PBI. Different from CCI and PBI that both ob
 tain run-time information through sampling\, LXR obtains run-time informat
 ion through hardware support that maintains recent execution history with 
 negligible overhead. I will conclude the talk by discussing other research
  in my group that tackles concurrency bugs and performance bugs.
LOCATION:BC 420 https://plan.epfl.ch/?room==BC%20420
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR