Date: Wed, 11 Nov 98 12:17:01 -0600 From: "Richard Seaman, Jr." <lists@tar.com> To: "Brian Feldman" <green@unixhelp.org> Cc: "current@freebsd.org" <current@FreeBSD.ORG> Subject: Re: RFSIGSHARE ready? Message-ID: <199811111817.MAA03748@ns.tar.com>
next in thread | raw e-mail | index | archive | help
On Wed, 11 Nov 1998 13:04:10 -0500 (EST), Brian Feldman wrote: >On Wed, 11 Nov 1998, Richard Seaman, Jr. wrote: > >> ==================BEGIN FORWARDED MESSAGE================== >> >From: "Richard Seaman, Jr." <lists@tar.com> >> >To: "Brian Feldman" <green@unixhelp.org> >> >Cc: "current@freebsd." <current@freebsd> >> >Date: Wed, 11 Nov 98 11:51:43 -0600 >> >Reply-To: "Richard Seaman, Jr." <lists@tar.com> >> >Subject: Re: RFSIGSHARE ready? >> > >> >> On Wed, 11 Nov 1998 12:23:46 -0500 (EST), Brian Feldman wrote: >> >> >In some programs, the program >> >getsto spinning in pthread_handle_create and CANNOT be killed, for some >> >reason or another. In other cases, tsleep (invoked from sigsuspend) is >> >sleeping forever, of course wakeup works but it's still in a while() loop >> >so it just continues. In this manner, both situations are unkillable, no >> >matter WHAT signal is sent (I sent all of them to test by the way) >> >> I found a similar problem when I first invoked a threaded program >> using linux threads. The problem went away when I moved >> THREAD_STACK_START_ADDRESS down to 0xe0000000. > >Creating an unkillable process? this is _NOT_ a good thing. >And should not be able to be accomplished under any circumstances. Tho, I >can think of now, ttywait (I believe) is one such. Actually it wasn't unkillable, it just hung in pthread_handle_create. FYI, I just tried recompiling with the stack size set to INITIAL_STACK_SIZE, instead of STACK_SIZE, and it also hangs in pthread_handle_create. I still think you have compatibility problems between the linux mmap with MAP_GROWSDOWN and the FreeBSD mmap. The stack in question here is the stack of the new threads, not the intial process. linuxthreads manages this stack, and I'm not convinced the FreeBSD kernel does, without additional coaxing. Also, FYI I just tried a make buildworld with your patches (before the newest ones you just sent), and it breaks in /usr/src/bin/ps/keyword.c in lines 161-163. Also, FYI I have had trouble applying the patches you have posted to list, since it appears that some of the lines are getting wrapped. I don't think that my mailer is doing this, possibly this is happening either in your mailer, or in the -current list remailer. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199811111817.MAA03748>