Re: panic on yesterday's -CURRENT: linux emulation and vm (lockmgr: locking against myself)
Alan Cox wrote: On Thu, Sep 25, 2003 at 11:15:57AM -0400, Robert Watson wrote: ... #6 0xc049f355 in vm_fault (map=0xc6fc1700, vaddr=0, fault_type=1 '\001', fault_flags=0) at /usr/src/sys/vm/vm_fault.c:219 #7 0xc04eddd9 in trap_pfault (frame=0xdd699b18, usermode=0, eva=0) at /usr/src/sys/i386/i386/trap.c:709 #8 0xc04eda50 in trap (frame= {tf_fs = -1070333928, tf_es = -1067384816, tf_ds = 16, tf_edi = 0, tf_esi = -1068054086, tf_ebp = -580281484, tf_isp = -580281532, tf_ebx = 441, tf_edx = -968258896, tf_ecx = 0, tf_eax = -968258896, tf_trapno = 12, tf_err = 0, tf_eip = -1070325008, tf_cs = 8, tf_eflags = 66178, tf_esp = 0, tf_ss = -1068146900}) at /usr/src/sys/i386/i386/trap.c:418 #9 0xc04dd9e8 in calltrap () at {standard input}:102 #10 0xc04adbde in vm_page_sleep_if_busy (m=0x1b9, also_m_busy=1, msg=0x0) at /usr/src/sys/vm/vm_page.c:441 #11 0xc04abcfb in vm_object_split (entry=0xc6eea8ac) at /usr/src/sys/vm/vm_object.c:1226 ... Based upon the above information, it looks like vm_object_split() followed a bogus vm page pointer. This is in frequently executed code. So, I would hypothesize a race condition or transient hardware error is responsible. When my recent amd64 and i386 pmap changes are replicated on all platforms that will enable me to introduce additional assertions on the vm object locking. That may reveal some unsynchronized vm object accesses that could lead to the above problem. Alan ___ FWIW, I got this same panic a few days ago, running portinstall shortly after gnome2 had fillled my fd table for some unknown reason, on 5.1-RELEASE-p5. Adam ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: mksnap_ffs, snapshot issues, again
Julian Elischer said: > Would it not be possible to make the snapshot file not appear in a > directory until it si finished? (I know that would be 'wierd' > but it would give a "guaranteed" solution.. > That sounds kinda neat as a compile time option or non-default (or perhaps even default) tunable if it were possible and not too difficult. The problem should likely be addressed given people might want to create snapshots around the same time they run find via the maintanence scripts. -- Adam - Migus Dot Org (http://www.migus.org) ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: make buildkernel hang with SCHED_ULE
Dr. Richard E. Hawkins said: > On Thu, Aug 14, 2003 at 10:49:42PM -0400, Adam Migus wrote: >> Andrew Gallatin wrote: > >> WRT the mime thing. My apologies. It never occured to me as >> everyone I >> know personally uses a "real" mail reader. I'd attached them >> simply to >> keep the scrolling down and allow order independant viewing. >> Thanks for >> the tip. I'll just read them in as plain text in the future. > > If it gives mail or mh problems, it doesn't work on *real* mail > readers. > > :) > > hawk, ignobly usin mutt > -- > Richard E. Hawkins, Asst. Prof. of Economics/"\ ASCII ribbon > campaign > [EMAIL PROTECTED] Smeal 178 (814) 375-4700 \ / against HTML > mail > These opinions will not be those of Xand postings. > Penn State until it pays my retainer. / \ > Well, I guess we're back to that whole, what is "real" thing. So if for the sake of argument, if "works with mime" were the qualification for a "real" mail reader and everyone stopped using mime, would all mail readers become not real? Gee, that could be problematic. There must be something else that makes a mail reader "real." Otherwise we'd better hope people keep using mime. If continued, I think this belongs on -chat. :-) -- Adam - Migus Dot Org (http://www.migus.org) ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to "[EMAIL PROTECTED]"
date/time, standards n' stuff question
Folks, First of all I'm not sure if this is the right list. If it isn't please accept my apologies and divert the thread to the right one so I'll know for future reference. I'm using the following little program to generate nano-second timestamps for performance testing: int main(int argc, char **argv) { struct timespec ts; if (clock_gettime(CLOCK_REALTIME, &ts) != 0) { perror("clock_gettime"); return (1); } printf("%d.%09lu\n", ts.tv_sec, ts.tv_nsec); return (0); } My question is would it be right/acceptable/agreeable to augment /usr/bin/date to include an option to generate nanosecond precision dates? I'm thinking this might make things icky with regards to standards, etc. but if it were an option and didn't affect the existing functionality would it really matter? Would anyone care? Does anyone like or object to the idea? My motivation for asking is simply to avoid having to distribute/use this program with the performance testing stuff as previously (before I cared about nano-second precision) I just used /usr/bin/date to generate the datestamps. -- Adam - Migus Dot Org (http://www.migus.org) ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: make buildkernel hang with SCHED_ULE
Andrew Gallatin wrote: Adam Migus writes: > Folks, > While doing some performance analysis (doing make -j5 buildkernel) > on a set of 14 kernels I've hit one using the SCHED_ULE scheduler > that hangs. It happens every time but not necessarily in the same > place in the make. > <...> > The hardware is a dual Xeon box. The kernel is SMP w/ SCHED_ULE > instead of SCHED_4BSD, the options required for diskless and the > following two options: You have machdep.hlt_logical_cpus: 1 in your sysctl output. [BTW, lots of people read this mail via the web archives at http://docs.freebsd.org/cgi/getmsg.cgi?fetch=1073654+0+current/freebsd-current, where its impossible to view mime; it would be MUCH better for us if appended things like stack traces and sysctl output rather then scrambling them for no reason] SCHED_ULE is incompatible with halting logical CPUs. Something about it does't know the core isn't running, so it schedules a job there which never runs, and then it gets confused. When I boot a 1 CPU P4 with an SMP kernel and machdep.hlt_logical_cpus=1, it hangs before making it to multiuser mode.. Try setting machdep.hlt_logical_cpus=0 (via sysctl now, and in /boot/loader.conf so it doesn't happen again). Drew Andrew, WRT the mime thing. My apologies. It never occured to me as everyone I know personally uses a "real" mail reader. I'd attached them simply to keep the scrolling down and allow order independant viewing. Thanks for the tip. I'll just read them in as plain text in the future. WRT the sysctl value. Thanks for the tip. Is this to be considered a bug in SCHED_ULE? If the default is hlt_logical_cpus=1 I would think the scheduler should be able to handle it or deal with it appropriately. Perhaps ignoring the value, setting it to 0 internally or even just putting a warning message on boot? After all, not everyone RTFM's. :-) Thanks again, -- Adam - Migus Dot Org (http://www.migus.org) ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: make buildkernel hang with SCHED_ULE
Bruce M Simpson said: > On Mon, Aug 11, 2003 at 11:09:44PM -0400, Adam Migus wrote: >> The hardware is a dual Xeon box. The kernel is SMP w/ SCHED_ULE >> instead of SCHED_4BSD, the options required for diskless and the >> following two options: >> >> options CPU_FASTER_5X86_FPU >> options CPU_UPGRADE_HW_CACHE > > Why are you using these options? they shouldn't be required for a > Xeon. > > Are you able to break in to DDB to run a ps to inspect the process > table > via the debugger key or NMI? > > BMS > Bruce, I also do testing on a dual Althon and honestly didn't bother to research whether they'd have any affect. Could/would they be causing a problem? I'll recompile without them and try again at any rate. As of now I haven't tried anything. I'd image I can break to DDB as the machine is responsive when it happens. Can you give me any helpful hints on what to look for and/or how to diagnose? -- Adam - Migus Dot Org (http://www.migus.org) ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: Does linux-sun-jdk_1.4.2 work?
Adam said: > Eh, there's no need to flaming here. I never said > maintaining Java on > FreeBSD was easy. I said that Java support on FreeBSD > is dodgy, which is > really a well-known public fact, for anyone that's ever > done any reading > on the subject. In fact, when the core developers give > a public > interview, this is frequently a topic of discussion. I was addressing the fact that you took it upon yourself to _tell_ the mailing-list readers that the port was released to early. You aren't the maintainer of this or any other port. Therefor unless you stick an "in my opinion" in there you kind of sound like an ass... I made no comments about the reliability of Java on FreeBSD although I have read there are issues (thanks for assuming I didn't) but I use it routinely without problems. > Perhaps you should try working with Java 1.4.x on > FreeBSD before you > assume something about me that's highly inaccurate. I > think you'll find > very quickly that it doesn't work nicely unless the > process is running > as root. My apologies for being highly inaccurate. Why don't you tell me _why_ Java 1.4.x only works nicely on _your_ system when running as root? Hint: ktrace is your friend. -- Adam Migus - Research Scientist Network Associates Laboratories (http://www.nailabs.com) ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: Does linux-sun-jdk_1.4.2 work?
Adam said: > On Mon, 2003-07-21 at 08:55, Yamada Ken Takeshi wrote: >> I have the followinacquirer and cannot compile jdk14 >> on FreeBSD-current, July 20th. cvsup'ed with the >> latest >> linux_base. >> >> Is is unique to me alone? >> >> tyd3# /usr/local/linux-sun-jdk1.4.2/bin/java -version >> # >> # HotSpot Virtual Machine Error, Internal Error >> # Please report this error at >> # http://java.sun.com/cgi-bin/bugreport.cgi >> # >> # Java VM: Java HotSpot(TM) Client VM (1.4.2-b28 >> mixed mode) >> # >> # Error ID: 4F533F4C494E55580E43505001C9 >> # >> >> Heap at VM Abort: >> Heap > > It happens to me too. Looks like this port was released > too quickly. > Java on FreeBSD is always a bit dodgy, it seems. > > -- > Adam <[EMAIL PROTECTED]> > Actually it's trying to aquire a lock on a file in the linux proc filesystem which is most likely failing because you have the Linux proc file system mounted read-only. Try mounting it read-write... Hm... [EMAIL PROTECTED]:ttyp0:amigus$ grep -ci [EMAIL PROTECTED] /src/freebsd/ports/INDEX 0 [EMAIL PROTECTED]:ttyp0:amigus$ Perhaps you should try writing a port yourself before you judge others on whether or not their ports are released too early or not. When you get your foot out of your mouth also try using your full name in your email address too so people can use it when calling you a pompous ass. I wouldn't want people getting us confused... Regards, -- Adam Migus - Research Scientist Network Associates Laboratories (http://www.nailabs.com) ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: 5-STABLE Roadmap
> - Benchmarks and performance testing - Having a source of > reliable and >useful benchmarks is essential to identifying performance > problems and guarding against performance regressions. A > 'performance team' that is made up of people and resources > for formulating, developing, and executing benchmark tests > should be put into place soon. >Comparisons should be made against both FreeBSD 4.x and > Linux 2.4.x. Tests to consider are: > - the classic 'worldstone' > - webstone - /usr/ports/www/webstone > - Fstress - http://www.cs.duke.edu/ari/fstress > - ApacheBench - /usr/ports/www/p5-ApacheBench > - netperf - /usr/ports/benchmarks/netperf I don't wanna argue about which benchmark is better I just wanna sign up for the team. I do performance testing for the TrustedBSD Project. I have employer supplied hardware and time slated to perf testing. While I can't get paid to test non TrustedBSD testing, I can use the hardware (on my own time) to test. I'm more than happy to do that. -- Adam Migus - Research Scientist Network Associates Laboratories (http://www.nailabs.com) TrustedBSD (http://www.trustedbsd.org) FreeBSD (http://www.freebsd.org) To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: usbd patch (MAXUSBDEV no more)
Joe, funny you should mention. I actually tried posting this patch three times. The first two times I actually said in the (more elaborate) explanation of the patch that I as well am moving toward devd. So, I see your point and it's why I didn't bother spending more than 30 minutes doing this patch. I'd like to see usbd go away. I'd like to help devd take it's place. Adam On Thu, Feb 13, 2003 at 05:01:28PM +, Josef Karthauser wrote: > On Thu, Feb 13, 2003 at 11:02:29AM -0500, Adam Migus wrote: > > This patch makes usbd track a dynamic number of devices using a list > > instead of the static array of 4 devices. It's implemented as a list > > but it's very easy to change. > > Thanks, although I'm not clear as to the future of usbd. With the > advent of devd most of usbd's functionality will probably me implemented > over there instead. I've not spent a great deal of time looking at devd > yet to be sure. > > Joe > -- > Josef Karthauser ([EMAIL PROTECTED])http://www.josef-k.net/ > FreeBSD (cvs meister, admin and hacker) http://www.uk.FreeBSD.org/ > Physics Particle Theory (student) http://www.pact.cpes.sussex.ac.uk/ > An eclectic mix of fact and theory. = To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: usbd patch (MAXUSBDEV no more)
Me too, so if it does require one someone please tell me. I obviously read the code and from my understanding no locking is fine but I could be wrong. On Thu, Feb 13, 2003 at 11:13:31AM -0500, Andrew R. Reiter wrote: > On Thu, 13 Feb 2003, Adam Migus wrote: > > :This patch makes usbd track a dynamic number of devices using a list > :instead of the static array of 4 devices. It's implemented as a list > :but it's very easy to change. > > Does this list want a lock to protect it? I am unfamiliar with usb > locking at the moment, so ignore if stupid. > > Cheers, > Andrew > > -- > Andrew R. Reiter > [EMAIL PROTECTED] > [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
usbd patch (MAXUSBDEV no more)
This patch makes usbd track a dynamic number of devices using a list instead of the static array of 4 devices. It's implemented as a list but it's very easy to change. -- Adam Migus - Research Scientist Network Associates Laboratories (http://www.nailabs.com) TrustedBSD (http://www.trustedbsd.org) FreeBSD (http://www.freebsd.org) --- ../../../../freebsd/src/usr.sbin/usbd/usbd.cThu Jan 23 05:36:35 2003 +++ usbd.c Tue Feb 11 21:21:55 2003 @@ -74,11 +74,6 @@ */ #define USBDEV "/dev/usb" -/* Maximum number of USB busses expected to be in a system - * XXX should be replaced by dynamic allocation. - */ -#define MAXUSBDEV 4 - /* Sometimes a device does not respond in time for interrupt * driven explore to find it. Therefore we run an exploration * at regular intervals to catch those. @@ -95,9 +90,13 @@ char *configfile = CONFIGFILE; /* name of configuration file */ -char *devs[MAXUSBDEV]; /* device names */ -int fds[MAXUSBDEV];/* file descriptors for USBDEV\d+ */ -int ndevs = 0; /* number of entries in fds / devs */ +struct usb_dev { + char *ud_name; + int ud_fd; + LIST_ENTRY(usb_dev) ud_list; +}; +LIST_HEAD(usb_dev_list, usb_dev) _usb_devs; +struct usb_dev_list *uds = &_usb_devs; int fd = -1; /* file descriptor for USBDEV */ int lineno; @@ -758,6 +757,7 @@ pid_t pid; struct sigaction ign, intact, quitact; sigset_t newsigblock, oldsigblock; + struct usb_dev *ud; int status; int i; @@ -789,8 +789,9 @@ /* child here */ /* close all open file handles for USBDEV\d* devices */ - for (i = 0; i < ndevs; i++) - close(fds[i]); /* USBDEV\d+ */ + LIST_FOREACH(ud, uds, ud_list) { + close(ud->ud_fd); /* USBDEV\d+ */ + } close(fd); /* USBDEV */ /* Restore original signal dispositions and exec the command. */ @@ -936,6 +937,9 @@ fd_set r,w; int itimeout = TIMEOUT; /* timeout for select */ struct timeval tv; + struct usb_dev *ud = NULL, *ud0 = NULL; + int fds; + LIST_INIT(uds); if (modfind(USB_UHUB) < 0) { if (kldload(USB_KLD) < 0 || modfind(USB_UHUB) < 0) { @@ -960,8 +964,18 @@ explore_once = 1; break; case 'f': - if (ndevs < MAXUSBDEV) - devs[ndevs++] = optarg; + ud0 = ud; + ud = (struct usb_dev *)malloc(sizeof(*ud)); + if (ud == NULL) { + fprintf(stderr, + "can't alloc space for %s\n", buf); + return 1; + } + ud->ud_name = optarg; + if (ud0 != NULL && !(LIST_EMPTY(uds))) + LIST_INSERT_AFTER(ud0, ud, ud_list); + else + LIST_INSERT_HEAD(uds, ud, ud_list); break; case 'n': handle_events = 0; @@ -981,51 +995,64 @@ argv += optind; maxfd = 0; - if (ndevs == 0) { + if (LIST_EMPTY(uds)) { /* open all the USBDEVS\d+ devices */ - for (i = 0; i < MAXUSBDEV; i++) { + for (i = 0;; i++) { sprintf(buf, "%s%d", USBDEV, i); - fds[ndevs] = open(buf, O_RDWR); - if (fds[ndevs] >= 0) { - devs[ndevs] = strdup(buf); - if (devs[ndevs] == NULL) { - fprintf(stderr, "strdup returned NULL\n"); - return 1; - } - if (verbose) - printf("%s: opened %s\n", - __progname, devs[ndevs]); - if (fds[ndevs] > maxfd) - maxfd = fds[ndevs]; - ndevs++; - } else if (errno != ENXIO && errno != ENOENT) { - /* there was an error, on a device that does -* exist (device is configured) -*/ - fprintf(stderr, "%s: Could not open %s, %s\n", - __progname, buf, strerror(errno)); - exit(1); +
Re: where is MAKEDEV
> On Mon, Jan 27, 2003 at 09:14:36PM +0100, FredBriand wrote: >> everything seems OK, but my sound card. In the handbook they >> say I must use the MAKEDEV script (as in Linux) but I can't >> find it on my disk. > > freebsd 5.0 is using devfs(5). so device inodes are created > automatically. > > for your soundcard be sure to enable pcm(4) in your kernel. > > for info on compiling a new kernel see the freebsd handbook: > > http://www.freebsd.org/doc/en_US.ISO8859-1/books/handbook/kernelconfig.html > It may be easier to just load the module. Try adding it to your /boot/loader.conf. For a list of available sound modules try grep ^snd /boot/defaults/loader.conf. If you don't know what your card is just type "kldload snd" That will load all sound drivers and "dmesg" will show you which one actually supports your card. Once you know, put that one in your /boot/loader.conf. If your driver requires pcm, the loader will load it as a dependancy. Adam -- Adam Migus - Research Scientist Network Associates Laboratories (http://www.nailabs.com) TrustedBSD (http://www.trustedbsd.org) FreeBSD (http://www.freebsd.org) To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: Force 1 CPU on SMP kernel?
> > > On Tue, 28 Jan 2003, Adam Migus wrote: > >> Yes, just to be more explicit. What I want to do is get >> the >> 'real' difference a second processor makes. ie. If I run >> GENERIC I'm not incurring the overhead associated with >> SMP/APIC. >> I want to use the same kernel making as small a change as >> nessessary to have the machine start/use just one CPU. > > I'm not aware of any hacks to disable APs through > loader.conf. > If you're trying to benchmark the system's performance with > 1 > vs N CPUs, you're best off pulling all but the BSP CPU out > of > the machine. For such a benchmark to be accurate, you > shouldn't make any software changes (kernel, daemons or > startup configs). I tend to agree but things like warrenties and flakey boards preclude removal of CPU(s). All the same I think the ability to turn on/off AP's at boot would be useful in general. -- Adam Migus - Research Scientist Network Associates Laboratories (http://www.nailabs.com) TrustedBSD (http://www.trustedbsd.org) FreeBSD (http://www.freebsd.org) To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: Force 1 CPU on SMP kernel?
> * De: Odhiambo Washington <[EMAIL PROTECTED]> [ Data: > 2003-01-28 ] > [ Subjecte: Re: Force 1 CPU on SMP kernel? ] >> * Ceri Davies <[EMAIL PROTECTED]> [20030128 12:50]: >> wrote: >> > On Mon, Jan 27, 2003 at 11:33:02PM -0500, Adam Migus >> wrote: >> > > Hi, >> > > I'm doing some performance analysis on FreeBSD. I would >> like to be able to make an SMP kernel use only one CPU. >> Of course I realize running GENERIC would only use 1 CPU >> but the concern is that the overhead assoicated with SMP >> will skew the statistics. So my question is what would >> be the 'best' way to make an SMP kernel only start/use >> CPU0? >> > >> > Yank the others ? >> > >> >> What about building two kernels, one SMP, the other non-SMP, >> then just boot wheichever kernel you want? Doesn't it server >> same purpose? Better than Ceri's idea ;) > > You're missing the point. He wants to boot only the bootstrap > processor, none of the APs, but still be running an SMP > kernel. I think checking how much overhead "SMP" and > "APIC_IO" add is a noble goal, and throwing real SMP into the > mix would be dumb. Yes, just to be more explicit. What I want to do is get the 'real' difference a second processor makes. ie. If I run GENERIC I'm not incurring the overhead associated with SMP/APIC. I want to use the same kernel making as small a change as nessessary to have the machine start/use just one CPU. To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Force 1 CPU on SMP kernel?
Hi, I'm doing some performance analysis on FreeBSD. I would like to be able to make an SMP kernel use only one CPU. Of course I realize running GENERIC would only use 1 CPU but the concern is that the overhead assoicated with SMP will skew the statistics. So my question is what would be the 'best' way to make an SMP kernel only start/use CPU0? -- Adam Migus - Research Scientist Network Associates Laboratories (http://www.nailabs.com) TrustedBSD (http://www.trustedbsd.org) FreeBSD (http://www.freebsd.org) To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: 80386 out of GENERIC
I have a 486 running as a firewall. I honestly didn't even think about trying 5.0 on that thing, forget a 386. So you don't get upset when Windows XP requires a Pentium two million with a bejigabyte of RAM but you complain when 5.0 Won't run on a 386? Adam > In message: <[EMAIL PROTECTED]> > [EMAIL PROTECTED] writes: > : >I don't feel this is a good decision. (I still have a 486, > act as a : >small server and a 286 witch is in storage) This > basally means that : >any one who doesn't have the latest > processor can't install FreeBSD. : > : No it doesn't mean that. > : > : FreeBSD still runs on all 386 family CPUs, the only > difference is that : if you want to run it on a 80386 you need > to enable an option in : your kernel config file. > : > : It will out of the box run on 486 and anything later. > > One problem with most 386 boxes is that they have very little > memory. sysinstall is a big, bloated pig dog these days that > takes more RAM than most 386 boxes have. This is true also > for many 486 boxes too. So even if 386 stuff were in the > default kernel, you'd likely have other issues in making > sysinstall work and have to do custom > hacking... > > Warner > > To Unsubscribe: send mail to [EMAIL PROTECTED] > with "unsubscribe freebsd-current" in the body of the message -- Adam Migus Network Associates Laboratories (http://www.nailabs.com) TrustedBSD (http://www.trustedbsd.org) | The Power To Secure To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
nullfs/unionfs
Recently I've been doing some twisted things with nullfs/unionfs. I've managed to contain myself to using read-only nullfs, but, the whole point of the unionfs usage is to isolate changes (ie. I need read-write). Anyway I've hit a VFS bug that upon reading current@, appears as though it might be related to the bug involving vnode backed md's. I just discovered the bug and know no more about it than what has been discussed in the related post. The only thing I can add is that when it happens to me, a shutdown of the system results in a panic due to an invalid ref count on the problematic vnode. The count goes negative. When the system comes back up, an fsck shows every inode in sequence from the bad one as unreferenced. I read the man pages for nullfs/unionfs. They mentioned that the code is orphaned and buggy. So, I blamed it, and entertained the idea of claiming ownership, now, or in future. Right now I sit somewhere in the middle of the (IMHO) rather steep VFS learning curve. Therefor even if I did sit down and attempt to track it I'd likely have limited success and if I found the bug, it'd probably be nowhere near nullfs/unionfs... In summary, two things. First WRT the nullfs/unionfs. I'd like to use it as I've tried to thus far. So I have an intrinsic motivation to get it fixed. Seems as though others do as well (assuming it's the same bug). So I think I'll have a go at it, but won't promise results as I'm not a VFS guru. Second, if anyone expresses interest in ownership of the nullfs/unionfs code, I'd like to work with them. If no one does and someone wants me to, I may, but at a later date when I'm more comfortable doing so. Third, if someone else attempts to find/fix the bug, consider me an interested. -- Adam Migus ([EMAIL PROTECTED]) ([EMAIL PROTECTED]) TrustedBSD (http://www.trustedbsd.org) | The Power To Secure To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message