Journal: Folding@Home

I've been working on a particularly vexing problem for a few days now, and I just realized that posting it here as my first Slashdot journal entry *might* just catch the eye of someone with an answer.

I'm trying to get the Folding@Home client to run on OpenBSD under Linux emulation. Seeing as how it already runs under FreeBSD with Linux emulation, it should be easy, right?

Of course not. Right now, it runs on one of my OBSD 3.3 boxes, but not the other, and the one it does run on takes some manual intervention. For the one that does run, I took a fresh install of 3.3, laid down Linux emulation from /usr/ports/emulators/redhat/base (version 6.2p7), and downloaded the FAH3Console-Linux.exe binary from the F@H people. I then had to run "elf2olf -o linux FAH3Console-Linux.ex" to properly tag the binary as Linux, otherwise the system thought it was svr4. At that point, it ran...but the instant it tried to go to the web to download a core binary, it coredumped. When I snagged those core binaries from a working Linux box and elf2olf'd them, though, it worked like a charm.

On my other box, though, things haven't been so smooth. I repeated the setup, with my version 6.2p5 Linux emulation, and while it ran, it never finished a work unit. I comletely uninstalled that version of Linux emulation and laid down the same version as on the working box...only to have the same inability to complete a frame. Nothing I've done, or had suggested, has been able to fix this.

Meanwhile, in trying to figure out why core binaries can't be downloaded, I've found that a SIGSEV occurs when the program tries to open /emul/linux/lib/libnss_files.so.2, as follows:

9484 FAH3Console-Linu CALL open(1208875670,0,-817902964)
9484 FAH3Console-Linu NAMI "/emul/linux/lib/libnss_files.so.2"
9484 FAH3Console-Linu NAMI "/emul/linux"
9484 FAH3Console-Linu NAMI "/emul/linux/lib/libnss_files.so.2"
9484 FAH3Console-Linu RET open 7
9484 FAH3Console-Linu CALL fstat(7,-817903660)
9484 FAH3Console-Linu RET fstat 0
9484 FAH3Console-Linu PSIG SIGSEGV SIG_DFL code 2 addr=0xcf3fb824 trapno=2
9484 FAH3Console-Linu PSIG SIGSEGV SIG_DFL code 0 addr=0x0 trapno=0
9484 FAH3Console-Linu NAMI "FAH3Console-Linu.core"
19926 FAH3Console-Linu PSIG SIGSYS caught handler=0x80a2048 mask=0x6ffef6ef
19926 FAH3Console-Linu RET poll -1 errno 4 Interrupted system call
19926 FAH3Console-Linu CALL sigreturn(137227512)
19926 FAH3Console-Linu RET sigreturn JUSTRETURN
19926 FAH3Console-Linu CALL getppid
19926 FAH3Console-Linu RET getppid 9735/0x2607
19926 FAH3Console-Linu CALL wait4(-1,137227636,-2147483647,0)
19926 FAH3Console-Linu RET wait4 9484/0x250c
19926 FAH3Console-Linu CALL kill(31399,11)
31399 FAH3Console-Linu PSIG SIGSEGV SIG_DFL code 2 addr=0x0 trapno=0
31399 FAH3Console-Linu NAMI "FAH3Console-Linu.core"
19926 FAH3Console-Linu RET kill 0
19926 FAH3Console-Linu CALL kill(9735,11)
9735 FAH3Console-Linu PSIG SIGSEGV SIG_DFL code 2 addr=0x0 trapno=0
9735 FAH3Console-Linu NAMI "FAH3Console-Linu.core"
19926 FAH3Console-Linu RET kill 0
19926 FAH3Console-Linu CALL exit(0)

The problem is, I've not NO idea what this library is all about, or how to debug a problem with it. Google, the OpenBSD archives, etc. are completely unhelpful.

Thus this journal entry. I figure, if nothing else, it helps ease my mental anguish over not being able to figure this out to write about it; and meanwhile, perhaps someone knowledgeable will stumble across it. You never know; after all, this is Slashdot!