Re: RegressionTest Results
On 11/26/2010 01:16 AM, Eric Bakan wrote: Here are the results of running the pcca-test framework. Thanks Eric. I looked at the first two [ 1/ 1] /aio.h/t_aio: Assertion failed: (aio_read(iocb) != -1), function main, file t_aio.c, line 57. [ 1/ 1] /aio.h/t_aio_suspend: Assertion failed: (lio_listio(LIO_WAIT, iocblist, NAIO, NULL) == 0), function main, file t_aio_suspend.c, line 71. aio is not enabled in the default kernel so not much to see here. It's returning ENOSYS. ---CUT--- [ 1/ 1] /fnmatch.h/t_fnmatch: Assertion failed: (rv == FNM_NOMATCH), function main, file t_fnmatch.c, line 40. [ 1/ 1] /fnmatch.h/t_fnmatch2: passed The test case that is failing is rv = fnmatch(\\, \\, 0); assert(rv == FNM_NOMATCH); Now without FNM_NOESCAPE fnmatch says it will treat the backslash as an escape and use the next character in the string to try and match with. It does not say it will do anything special when the next character is null. So so it seems that pattern should not match. Just like this one that does not return a match : fnmatch(\\string, \\string, 0); This will incorrectly match : fnmatch(string\\, string\\, 0); This patch changes the trailing backslashes from matching. http://leaf.dragonflybsd.org/~dylan/0001-fnmatch-Fix-edge-case-with-trailing-blackslash.patch Dylan
Re: No package installation method works
On 10/20/2010 11:30 PM, Torbjorn Granlund wrote: Someone replied privately: The problem here is that you are using make instead of bmake (which needs to be used for pkgsrc things). Ah, this indeed work. I'll try to remember this when I, in some months, need to install some additional pacage. Thanks to all who replied! What if make gave a 'hint' when run in /usr/pkgsrc. Something like # make make called in /usr/pkgsrc/www/zope, did you mean bmake? make: no target to make. diff --git a/usr.bin/make/main.c b/usr.bin/make/main.c index 5294f03..2f744a9 100644 --- a/usr.bin/make/main.c +++ b/usr.bin/make/main.c @@ -887,6 +887,7 @@ main(int argc, char **argv) charcurdir[MAXPATHLEN]; /* startup directory */ charobjdir[MAXPATHLEN]; /* where we chdir'ed to */ const char *make_flags; + const char *bmake_hint_path = /usr/pkgsrc; /** * This section initializes stuff that require no input. @@ -953,6 +954,8 @@ main(int argc, char **argv) Var_SetGlobal(.CURDIR, curdir); Var_SetGlobal(.OBJDIR, objdir); + if(strspn(bmake_hint_path,curdir) == strlen(bmake_hint_path)) + fprintf(stderr,make called in %s, did you mean bmake?\n,curdir); /* * Set up the .TARGETS variable to contain the list of targets to be * created. If none specified, make the variable empty -- the parser
Re: USB image
On 09/26/2010 01:32 PM, tron wrote: If I want to copy the DF USB image onto my USB stick and I am working under WinXP, can I open the archive with 7zip and just copy the resulting files/folders to the stick or will I need a special app for installing the image to the USB stick? (Sorry if this sounds dumb, but I am only used to the world of Windows and the handbook on DF page only discusses installation with CD.) You should be able to use Image Writter https://launchpad.net/win32-image-writer Dylan
Re: Utility to list /dev/nodes serno's
On 08/06/2010 01:27 PM, Matthew Dillon wrote: :Hi people, : :is there a way to easily list all disks and their associated serno's ? :Something like 'blkid' utility of Linux, if you happen to know it. :I could happily hack something like that, if we lack it. : : :Cheers, :Stathis There isn't, and that would be cool. It is fairly easy to match up device numbers from devfs. You can use sysctl kern.disks output to get a list of disk devices, then you can scan /dev and pick those base names out and stat them, and you can scan /dev/serno and stat those babies and match up the st_rdev's with the ones from /dev to getting related serial numbers. If we wanted to get more involved we could add an ioctl() to retrieve the serial number (if available), but it can definitely be scripted right now without that. -Matt Matthew Dillon dil...@backplane.com I started a utility a while back going down a different path, I have now changed it to use the sysctl call like Matt pointed out. This is my first attempt to contribute a utility to DragonFly (or any project) so I'm looking for feedback. I have more error checking to add. I was not sure the best place for the code, so I created a new empty branch [devserno] on my leaf account with the code. http://gitweb.dragonflybsd.org/~dylan/dragonfly.git/commit/refs/heads/devserno And a tarball http://leaf.dragonflybsd.org/~dylan/devserno.tar.gz The program reads the rdev's from all files in /dev/serno (excluding any with a .) then reads the sysctl kern.disk and prints out the matches. This is the output from my system : $ ./devserno DEVICE SERNO /dev/ad205008023 /dev/ad4JPB530HN256PBB /dev/ad6WD-WXEZ07T50371 Regards, Dylan
Re: swapcache Setup
On 05/24/2010 01:42 PM, Matthew Dillon wrote: :Hello All, : : I added a How-To for setting up swapcache from my notes. : :http://www.dragonflybsd.org/docs/howtos/swapcache/ : :Thanks, :Dylan That looks pretty good. The swapcache manual page has some extensive recommendations (but not the step-by-step that you put together for docs), the most important of which being that you might want to configure only 32G of swap on the 40G SSD instead of using all 40G. For a manufacturer-fresh SSD configuring only 32G out of the 40G leaves 8G which the SSD firmware will use to enhance its wear-leveling algorithms, improving the overall life of the SSD. Here are some suggestions: * When using a 40G SSD as an example configure only 32G of swap. * There is no need to newfs a swap partition. * Someone configuring swapcache fresh with no prior experience should start by only caching meta-data. i.e. data_enable=0. -Matt Matthew Dillon dil...@backplane.com Matt, Thanks for the info, I will update the page to include your suggestions. Dylan
swapcache Setup
Hello All, I added a How-To for setting up swapcache from my notes. http://www.dragonflybsd.org/docs/howtos/swapcache/ Thanks, Dylan
HAMMER:WARNING: Missing inode for dirent
I am getting these errors when I run undo update.log HAMMER: WARNING: Missing inode for dirent update.log@@0x0001166e7da0 obj_id = 000103bca45e, asof=0001166e7da0, lo= HAMMER: WARNING: Missing inode for dirent update.log@@0x000116a785c0 obj_id = 000103bca45e, asof=000116a785c0, lo= The system was just updated. DragonFly backup_a.gasdasoftware.com 2.7-DEVELOPMENT DragonFly v2.7.2.75.g28b766-DEVELOPMENT #6: Tue Apr 27 21:20:44 PDT 2010 r...@backup_a.gasdasoftware.com:/usr/obj/usr/src/sys/GENERIC i386 Should I worry? Thanks, Dylan
Re: Insufficients buffers for rebalance
Matthew Dillon wrote:
Don't bother trying to rebalance. It won't actually hurt the filesystem
much to not rebalance. The rebalancer needs a lot of buffer cache
buffers to operate at the moment and there's no easy solution other
then to add memory.
-Matt
Matt,
I get this with even 512MB of ram, as it does not seem to be something
we need people to worry about can we subdue the message unless
vfs.hammer.debug.general is 0x0080??
0x0080 might not been the correct one to use, but seemed to be popular,
is there a list of reasons for each one used in debug_general?
Attached is a patch.
Thanks,
Dylan
From 52862807299d79f0253d428a0b1529ad63e87bf3 Mon Sep 17 00:00:00 2001
From: Dylan Reinhold dy...@ocnetworking.com
Date: Tue, 27 Apr 2010 22:04:21 -0700
Subject: [PATCH] HAMMER: Suppress rebalance buffer message.
Only show the message if debug.general is 0x0080
---
sys/vfs/hammer/hammer_ioctl.c |6 --
1 files changed, 4 insertions(+), 2 deletions(-)
diff --git a/sys/vfs/hammer/hammer_ioctl.c b/sys/vfs/hammer/hammer_ioctl.c
index f8fe994..dc7eb90 100644
--- a/sys/vfs/hammer/hammer_ioctl.c
+++ b/sys/vfs/hammer/hammer_ioctl.c
@@ -94,9 +94,11 @@ hammer_ioctl(hammer_inode_t ip, u_long com, caddr_t data, int fflag,
* little memory will not be able to do it.
*/
if (error == 0 nbuf HAMMER_REBALANCE_MIN_BUFS) {
- kprintf(hammer: System has insufficient buffers
-to rebalance the tree. nbuf %d\n,
+ if (hammer_debug_general 0x0080) {
+kprintf(hammer: System has insufficient
+buffers to rebalance the tree. nbuf %d\n,
HAMMER_REBALANCE_MIN_BUFS);
+ }
error = ENOSPC;
}
if (error == 0) {
--
1.6.4
USB Wireless
I'm trying to get my DF machine to connect via this Linksys Wireless-G USB device. The system sees it a generic USB device ugen0: Broadcom Linksys Wireless-G USB Network Adapter with SpeedBooster, class 2/0, rev 2.00/0.06, addr 2 on uhub0 Any ideas on how (and if) I can get this to work. Thanks, Dylan
Re: Determining Running Applications on Specific CPU
This is the top output from a MP Box load averages: 2.29, 1.02, 0.41 up 0+00:05:03 21:33:47 39 processes: 39 running CPU0 states: 16.2% user, 0.0% nice, 53.0% system, 21.1% interrupt, 9.7% idle CPU1 states: 39.5% user, 0.0% nice, 23.5% system, 27.1% interrupt, 10.0% idle Mem: 13M Active, 81M Inact, 79M Wired, 20K Cache, 60M Buf, 322M Free Swap: 1024M Total, 1024M Free Hope this helps Dylan Archimedes Gaviola wrote: To Whom It May Concerned: Hello and good day! I would like to know if DragonFly is capable of determining running user applications on specific CPU? Let say for example, a system with 2 processors running on 3 different user applications. One application is running on the first CPU while the remaining 2 applications running on the 2nd CPU? While running many applications on a CPU, it will also determine how much CPU utilization percentage each running application is consuming? Thanks, Archimedes
Re: lpd is broken
Vincent Stemen wrote: Sorry guys. Another problem you may not want to propagate to the 2.0 release. lpd works for local printing but is broken for remote printing. .SNIP... Our printcaps: Client: lp|Line Printer:\ :sh:lp=:rm=10.0.0.201:rp=:\ :sd=/var/spool/output/lpd:\ :lf=/var/log/lpd-errs: ..SNIP... Vincent, I was able to reproduce your same results. It looks like lpd is being confused by the parameters in your printcap. Bug or not, if I remove the 'rp=' it works for me. It seems to have an issue where you set the remote printer to nothing, where it is set to lp by default if you leave this out. Dylan
Re: Problem Booting new System 1.12 live cd
Matthew Dillon wrote: Hmm. Well, the question is what is stopping it.. it could be the ATA driver. I'm assuming you don't have any mass storage plugged into the USB. One thing to try... let the boot proceed and give it a good 5 minutes to try to get past the xpt warnings. The xpt stuff will give up after a few minutes and there is an outside chance that the machine will be able to complete its boot. -Matt Matthew Dillon [EMAIL PROTECTED] Ok it looks like it was the state of the new ATA code at 1.12.1_RELEASE. I was able to install from a 1.8.1_RELEASE CD. Then I updated to head, the system was able to boot. Thanks, Dylan
Problem Booting new System 1.12 live cd
I have not used DragonFly in a long time (I think the last ver was 1.0). I just got a new system, so I wanted to see how DragonFly would play, but the live CD 1.12 (from Simon's site) wont boot with or without ACPI. It starts to boot and then keeps repeating this message intr 10 at 40001/4 hz, livelock limit engage! intr 10 at 19810/2 hz livelock removed. ... then a few of these show up *WARNING* waiting for the following device finish xpt: func 0xC01500E35 arg=0 The system is an Intel Quad Core 2.40 GHz on a Gigabyte GA-P35-S3G motherboard. Any ideas? Thanks, Dylan
