Date: Tue, 22 Jan 2008 04:49:59 +0000 From: Bruce M Simpson <bms@incunabulum.net> To: pluknet <pluknet@gmail.com> Cc: FreeBSD Current <freebsd-current@freebsd.org> Subject: Re: exmap port - shared memory profiler Message-ID: <479575F7.80605@incunabulum.net> In-Reply-To: <a31046fc0801202122o60a14b96g40cdaa9701738e84@mail.gmail.com> References: <478B89F6.1010703@incunabulum.net> <a31046fc0801151837k723cc58dp9d67b454befe0cb@mail.gmail.com> <479012F5.9050408@incunabulum.net> <a31046fc0801202122o60a14b96g40cdaa9701738e84@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
pluknet wrote: > I've slightly cleared exmap port to more working state. Now it > compiles (with also patched sysutils/pmap) and all binaries from this > port works (including gtk tool) on 7.0 (still not tested on 6.x and > HEAD). > Thanks for the help! I almost have it running on 6.3-RELEASE with these changes. I haven't tested with -CURRENT. Now it crashes due to finding some kind of inconsistency in the page map it has learned. Compiling with -O0 and then examining the core file suggests it's creating the FbsdSysInfo instance again. Exmap is still tied to being able to reach into Linux and grab physical page maps. I think the best way forward may be to adapt the pmap_helper.ko module and extend it to return the result of its call to pmap_extract(), rather than faking 'page cookies' as I have currently patched Exmap to do. There's no way for it to learn precisely which pages are mapped and resident for the same reason, and it really should. I have to crack on with other work, I may get around to this later on. thanks again BMS
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?479575F7.80605>