This is the mail archive of the systemtap@sourceware.org mailing list for the systemtap project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[Bug translator/10507] overhuge BSS for many-uprobe module


------- Additional Comments From fche at redhat dot com  2009-08-11 20:34 -------
This is an effect of this generated code:

#ifndef MULTIPLE_UPROBES
#define MULTIPLE_UPROBES 256
#endif
#ifndef MAXUPROBES
#define MAXUPROBES (MULTIPLE_UPROBES * 24743)
#endif
static struct stap_uprobe {
  union { struct uprobe up; struct uretprobe urp; };
  int spec_index;
} stap_uprobes [MAXUPROBES];


IOW, we preallocate 256 * #number-of-probes copies of struct {up|urp}, to
account for up to 256 concurrent processes.

Short of dynamically allocating the uprobe/uretprobe structs as threads
come & go (and correspondingly complicating the uprobe* -> stap_uprobe_specs[]
mapping), we could heuristically turn down that multiplier.

For example, we could guess that a script with thousands of probes is likely
to apply to a smaller number of concurrent processes.


-- 


http://sourceware.org/bugzilla/show_bug.cgi?id=10507

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]