This is the mail archive of the
mailing list for the systemtap project.
Re: [PATCH] kprobes for s390 architecture
- From: fche at redhat dot com (Frank Ch. Eigler)
- To: Mike Grundy <grundym at us dot ibm dot com>
- Cc: Heiko Carstens <heiko dot carstens at de dot ibm dot com>, Martin Schwidefsky <schwidefsky at de dot ibm dot com>, Jan Glauber <jan dot glauber at de dot ibm dot com>, systemtap at sources dot redhat dot com, dwilder at us dot ibm dot com
- Date: 10 Jul 2006 10:25:11 -0400
- Subject: Re: [PATCH] kprobes for s390 architecture
- References: <20060623150344.GL9446@osiris.boeblingen.de.ibm.com> <OF44DB398C.F7A51098-ON88257196.007CD277-88257196.007DC8F0@us.ibm.com> <20060623222106.GA25410@osiris.ibm.com> <20060624113641.GB10403@osiris.ibm.com> <1151421789.5390.65.camel@localhost> <20060628055857.GA9452@osiris.boeblingen.de.ibm.com> <20060707172333.GA12068@localhost.localdomain> <20060707172555.GA10452@osiris.ibm.com> <20060708185428.GA26129@localhost.localdomain>
> > How fast is this if you have to exchange several hundred instructions?
> I'll have to do some timing tests to measure how much overhead it
> puts on the system. It's a trade off of safety vs speed. [...]
No question, safety has to be paramount.
> The plus side is the swaps are only done on activation and
> deactivation, and don't have this kind of overhead during
> operations. [...]
Even though infrequent, we should nevertheless measure
activation/deactivation times, since it is not hard to insert and
remove thousands of kprobes in a single script. If those times are on
the order of milliseconds, and if batching them into one operation can
allow sharing of synchronization overhead, then we will need to adopt