Can't make 3.7-stable release
Hello! ...Same problem, again (it was already covered some time ago). When I run the last step in building a release (see http://www.openbsd.org/faq/faq5.html) , i.e. # make release I get a message informing me that /dev/svnd0a is full. This occurs while make is working with ramdiskC (exactly as the messages posted last July). Tried also on different hardware, same result. I've been struggling with this for a couple of weeks now. Three quick questions: A) Solution is the same as previously suggested (removing a non critical driver, such as axe, from ramdiskC) or has anything changed? B) After commenting out the axe driver, you have to start over and rebuild the kernel, right? Userland as well? (I'm asking since I'm working on a not-so-fast machine, the whole process takes quite some time...) C) Please don't flame--I'm just curious: In the mailing list archives, I noticed this sort of problem has been around since March (messages dated March 30). Why hasn't it yet been fixed? Thanks in advance for any suggestions. --Rob
Re: Can't make 3.7-stable release
[EMAIL PROTECTED] wrote: Hello! ...Same problem, again (it was already covered some time ago). When I run the last step in building a release (see http://www.openbsd.org/faq/faq5.html) , i.e. # make release I get a message informing me that /dev/svnd0a is full. This occurs while make is working with ramdiskC (exactly as the messages posted last July). C) Please don't flame--I'm just curious: In the mailing list archives, I noticed this sort of problem has been around since March (messages dated March 30). Why hasn't it yet been fixed? In the strictest of terms, a fix is impossible. Think about it a bit. The problem could be mitigated a bit by dropping a driver, but then its not the same release is it. And then there are the changes to the documentation, etc ... I'm not going to take time to go back and check, but I think I am the OP of the Mar 30 msg you refer to. Just do what I did, find a suitable work around (there are several) and get on with the show. Regards, Ray
Can't make 3.7-stable release (tries to exceed capacity of /dev/svnd0a?)
I've tried building an OpenBSD release from the 3.7-stable branch a
few times in the last few days, on two different i386 machines, and
both stopped in the same place. I'm following release(8) closely and
not trying to reuse /usr/obj, and dealing with new, clean, complete,
consistent checkouts of the code. The failure comes on step 4, (make
and validate the system release) during the make release command in
/usr/src/etc.
Here's a log of the failure. It seems to be working on RAMDISKC:
---
building standard compat library
ranlib libcompat.a
cc -Werror -Wall -Wstrict-prototypes -Wmissing-prototypes
-Wno-uninitialized -Wno-format -Wno-main -fno-stack-protector
-fno-builtin-printf -fno-builtin-log -Os -pipe -nostdinc -I.
-I/usr/src/sys/arch/i386/compile/RAMDISKC/../../../../arch
-I/usr/src/sys/arch/i386/compile/RAMDISKC/../../../.. -DSCSITERSE
-DAPM_NOPRINT -DI386_CPU -DI486_CPU -DI586_CPU -DI686_CPU -DSMALL_KERNEL
-DNO_PROPOLICE -DTIMEZONE=0 -DDST=0 -DFFS -DEXT2FS -DCD9660
-DMSDOSFS -DFIFO -DINET -DBOOT_CONFIG -DRAMDISK_HOOKS
-DMINIROOTSIZE=0xed8 -DPCIVERBOSE -D_KERNEL -Di386 -c swapbsd.c
sh /usr/src/sys/arch/i386/compile/RAMDISKC/../../../../conf/newvers.sh
cc -Werror -Wall -Wstrict-prototypes -Wmissing-prototypes
-Wno-uninitialized -Wno-format -Wno-main -fno-stack-protector
-fno-builtin-printf -fno-builtin-log -Os -pipe -nostdinc -I.
-I/usr/src/sys/arch/i386/compile/RAMDISKC/../../../../arch
-I/usr/src/sys/arch/i386/compile/RAMDISKC/../../../.. -DSCSITERSE
-DAPM_NOPRINT -DI386_CPU -DI486_CPU -DI586_CPU -DI686_CPU -DSMALL_KERNEL
-DNO_PROPOLICE -DTIMEZONE=0 -DDST=0 -DFFS -DEXT2FS -DCD9660
-DMSDOSFS -DFIFO -DINET -DBOOT_CONFIG -DRAMDISK_HOOKS
-DMINIROOTSIZE=0xed8 -DPCIVERBOSE -D_KERNEL -Di386 -c vers.c
rm -f bsd
ld -Ttext 0xD0100120 -e start -N -S -x -o bsd ${SYSTEM_OBJ} vers.o
textdatabss dec hex
1338427 1984260 281136 3603823 36fd6f
cp
/usr/src/distrib/i386/ramdiskC/../../../sys/arch/i386/compile/RAMDISKC/bsd
bsd
cc -DDEBUG -o rdsetroot
/usr/src/distrib/i386/ramdiskC/../../common/elfrdsetroot.c
cp bsd bsd.rd
/usr/src/distrib/i386/ramdiskC/obj/rdsetroot bsd.rd mr.fs
segment 0 rd_root_size_off = 0x14c5c0
rd_root_image_off = 0x14c5e0
rd_root_size val: 0x001DB000 (3800 blocks)
copying root image...
...copied 1945600 bytes
cp bsd.rd bsd.strip
strip bsd.strip
strip -R .comment bsd.strip
gzip -c9 bsd.strip bsd.gz
dd if=/dev/zero of=/var/tmp/image.19889 bs=10k count=144
144+0 records in
144+0 records out
1474560 bytes transferred in 0.031 secs (46081440 bytes/sec)
vnconfig -v -c svnd0 /var/tmp/image.19889
disklabel -w -r svnd0 floppy3
newfs -m 0 -o space -i 524288 -c 80 /dev/rsvnd0a
/dev/rsvnd0a: 2880 sectors in 80 cylinders of 2 tracks, 18 sectors
1.4MB in 1 cyl groups (80 c/g, 1.41MB/g, 32 i/g)
super-block backups (for fsck -b #) at:
32,
mount /dev/svnd0a /mnt
cp /usr/tmp/usr/mdec/boot /usr/src/distrib/i386/ramdiskC/obj/boot
strip /usr/src/distrib/i386/ramdiskC/obj/boot
strip -R .comment /usr/src/distrib/i386/ramdiskC/obj/boot
dd if=/usr/src/distrib/i386/ramdiskC/obj/boot of=/mnt/boot bs=512
75+1 records in
75+1 records out
38612 bytes transferred in 0.002 secs (17047241 bytes/sec)
dd if=bsd.gz of=/mnt/bsd bs=512
/mnt: write failed, file system is full
dd: /mnt/bsd: No space left on device
2720+1 records in
2720+0 records out
1392640 bytes transferred in 0.811 secs (1716327 bytes/sec)
*** Error code 1
Stop in /usr/src/distrib/i386/ramdiskC (line 30 of
/usr/src/distrib/i386/ramdiskC/../common/Makefile.inc).
*** Error code 1
Stop in /usr/src/distrib/i386.
*** Error code 1
Stop in /usr/src/distrib.
*** Error code 1
Stop in /usr/src/etc (line 376 of Makefile).
# df /mnt /
Filesystem 512-blocks Used Avail Capacity Mounted on
/dev/svnd0a 2815 280510 100%/mnt
/dev/wd0a 150652 72724 7039651%/
# ls -l /usr/obj/distrib/i386/ramdiskC/bsd.gz
-rw-r--r-- 1 root wsrc 1392775 Jul 11 01:30
/usr/obj/distrib/i386/ramdiskC/bsd.gz
# ls -l /usr/obj/distrib/i386/ramdiskC/boot
-r-xr-xr-x 1 root wsrc 38612 Jul 11 01:30
/usr/obj/distrib/i386/ramdiskC/boot
# uname -a
OpenBSD noc.hillconet.net 3.7 GENERIC#0 i386
(I was running the 3.7-stable GENERIC kernel both times I tried. The
userland, etc. builds and installs, too, just can't make the release.)
Here's a complete dmesg from one of the machines that it failed to build
on: (the other is pretty similar, and I believe both failures were
identical, though I didn't save the last log, because I figured I
probably did something wrong):
OpenBSD 3.7-stable (GENERIC) #0: Sun Jul 10 14:53:01 CDT 2005
[EMAIL PROTECTED]:/usr/src/sys/arch/i386/compile/GENERIC
cpu0: Intel Celeron (GenuineIntel 686-class, 128KB L2 cache) 468 MHz
cpu0:
FPU,V86,DE,PSE,TSC,MSR,PAE,MCE,CX8,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,MMX,FXS
R
real mem = 199794688 (195112K)
avail mem = 175558656 (171444K)
using 2464 buffers containing 10092544 bytes (9856K) of memory
mainbus0 (root)
bios0 at mainbus0: AT/286+(78)
Re: Can't make 3.7-stable release (tries to exceed capacity of /dev/svnd0a?)
On Mon, 11 Jul 2005, Adam Fabian wrote: I've tried building an OpenBSD release from the 3.7-stable branch a few times in the last few days, on two different i386 machines, and both stopped in the same place. I'm following release(8) closely and not trying to reuse /usr/obj, and dealing with new, clean, complete, consistent checkouts of the code. The failure comes on step 4, (make and validate the system release) during the make release command in /usr/src/etc. Here's a log of the failure. It seems to be working on RAMDISKC: Ummm - maybe I don't understand but, how can you make a RELEASE from STABLE? Isnt STABLE following the patch branch? And RELEASE is jsut that - what's on the CD? If i'm correct - then you can't do that. If I'm correct, think of it this way: RELEASE = what you buy. STABLE = is what you follow for security patches CURRENT = is what you run as a developer. It's kinda like a Beta of the next RELEASE.
Re: Can't make 3.7-stable release (tries to exceed capacity of /dev/svnd0a?)
On Mon, Jul 11, 2005 at 11:48:49AM -0500, Chris wrote: On Mon, 11 Jul 2005, Adam Fabian wrote: I've tried building an OpenBSD release from the 3.7-stable branch a Ummm - maybe I don't understand but, how can you make a RELEASE from STABLE? Isnt STABLE following the patch branch? And RELEASE is jsut that - what's on the CD? Perhaps it's not supported. I built a 3.5-stable release after about 16 updates, and it's a mite more convenient than doing a fresh install then applying 16 patches, though. ;) -- Adam Fabian [EMAIL PROTECTED]
Re: Can't make 3.7-stable release (tries to exceed capacity of /dev/svnd0a?)
Ummm - maybe I don't understand but, how can you make a RELEASE from STABLE? Isnt STABLE following the patch branch? And RELEASE is jsut that - what's on the CD? If i'm correct - then you can't do that. If I'm correct, think of it this way: RELEASE = what you buy. STABLE = is what you follow for security patches CURRENT = is what you run as a developer. It's kinda like a Beta of the next RELEASE. 3.7-release is what is on the cd. it is the result of building a release(8) from the source when 3.6 became 3.7-beta, solidified and changed to just 3.7. 3.7-stable is a branch of 3.7 made at the same time as 3.7-release was tagged. one can make a release(8) of the source in the branch at any date. the output would be similar base37.tgz, comp37.tgz, etc. sets to that of the 3.7-release, but with the minor -stable fixes in it. this is what the original poster is attempting. 3.7-current is the main development line of 3.7. one can make a release(8) of the source in the main development line at any date (assuming nothing is broken) and produce disk sets with new stuff from recent developments. this is what is done periodically to produce the snapshots that are available on the ftp mirrors. eventually 3.7-current turns into 3.8-beta, solidified, and then becomes 3.8. the tree is tagged as 3.8-release and branched as 3.8-stable and a release(8) is made with base38.tgz, comp38.tgz., etc. sets. these sets are put on a cd and distributed to the ftp mirrors. rinse, repeat every ~6 months.
Re: Can't make 3.7-stable release (tries to exceed capacity of /dev/svnd0a?)
On Mon, Jul 11, 2005 at 11:48:49AM -0500, Chris wrote: On Mon, 11 Jul 2005, Adam Fabian wrote: I've tried building an OpenBSD release from the 3.7-stable branch a few times in the last few days, on two different i386 machines, and both stopped in the same place. I'm following release(8) closely and not trying to reuse /usr/obj, and dealing with new, clean, complete, consistent checkouts of the code. The failure comes on step 4, (make and validate the system release) during the make release command in /usr/src/etc. Here's a log of the failure. It seems to be working on RAMDISKC: Ummm - maybe I don't understand but, how can you make a RELEASE from STABLE? Isnt STABLE following the patch branch? And RELEASE is jsut that - what's on the CD? If i'm correct - then you can't do that. If I'm correct, think of it this way: RELEASE = what you buy. STABLE = is what you follow for security patches CURRENT = is what you run as a developer. It's kinda like a Beta of the next RELEASE. This is one of those annoying re-use of words. You most certainly build a release(8) of -stable. or -current. or -release. :) There is an issue with RAMDISKC on -stable, it grew post-release. Either: 1) Remove a couple non-critical drivers from RAMDISKC or 2) Tell the system not to bother building RAMDISKC as part of the release. You can do this by removing references to ramdiskC from /usr/src/distrib/i386/Makefile. I started with attack 1, but decided 2 was easier after the third time I managed to miss one use of something, resulting in a bad build. Proper solution is 1, but I haven't had a chance to sit down and figure out what drivers are best to remove. Nick.
Re: Can't make 3.7-stable release (tries to exceed capacity of /dev/svnd0a?)
On 7/11/05, Chris [EMAIL PROTECTED] wrote: On Mon, 11 Jul 2005, Adam Fabian wrote: I've tried building an OpenBSD release from the 3.7-stable branch a few times in the last few days, on two different i386 machines, and both stopped in the same place. I'm following release(8) closely and not trying to reuse /usr/obj, and dealing with new, clean, complete, consistent checkouts of the code. The failure comes on step 4, (make and validate the system release) during the make release command in /usr/src/etc. Here's a log of the failure. It seems to be working on RAMDISKC: Ummm - maybe I don't understand but, how can you make a RELEASE from You aren't understanding. Read the release(8) manpage, it'll explain. It is fully supported. STABLE? Isnt STABLE following the patch branch? And RELEASE is jsut that - what's on the CD? If i'm correct - then you can't do that. If I'm correct, think of it this way: RELEASE = what you buy. STABLE = is what you follow for security patches CURRENT = is what you run as a developer. It's kinda like a Beta of the next RELEASE. Basically, you take the stable source branch, and build a CD release from that source. the release branch is what's on the cds, but building a release is just making a CD that's like the CD release, but from different sources if you so choose. The problem the OP is having was caused by one of the stable patches to the 3.7 kernel, making it just too big to fit on a floppy, which has been talked about in at least one previous thread. Jason
Re: Can't make 3.7-stable release (tries to exceed capacity of /dev/svnd0a?)
On Mon, Jul 11, 2005 at 01:24:01PM -0400, Jason Crawford wrote: from different sources if you so choose. The problem the OP is having was caused by one of the stable patches to the 3.7 kernel, making it just too big to fit on a floppy, which has been talked about in at least one previous thread. ::blushes.:: I suppose I should've found that. Thanks for the help, though! -- Adam Fabian [EMAIL PROTECTED]
Re: Can't make 3.7-stable release (tries to exceed capacity of /dev/svnd0a?)
Adam Fabian wrote: I've tried building an OpenBSD release from the 3.7-stable branch a few times in the last few days, on two different i386 machines, and both stopped in the same place. --8-- dd if=bsd.gz of=/mnt/bsd bs=512 /mnt: write failed, file system is full dd: /mnt/bsd: No space left on device --8-- # uname -a OpenBSD noc.hillconet.net 3.7 GENERIC#0 i386 (I was running the 3.7-stable GENERIC kernel both times I tried. The userland, etc. builds and installs, too, just can't make the release.) Here's a complete dmesg from one of the machines that it failed to build on: (the other is pretty similar, and I believe both failures were identical, though I didn't save the last log, because I figured I probably did something wrong): OpenBSD 3.7-stable (GENERIC) #0: Sun Jul 10 14:53:01 CDT 2005 [EMAIL PROTECTED]:/usr/src/sys/arch/i386/compile/GENERIC cpu0: Intel Celeron (GenuineIntel 686-class, 128KB L2 cache) 468 MHz cpu0: FPU,V86,DE,PSE,TSC,MSR,PAE,MCE,CX8,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,MMX,FXS R real mem = 199794688 (195112K) avail mem = 175558656 (171444K) This is really something for the archives. A problem description with really good info, and a dmesg too, even though it actually wasn't needed. Newbies, see how this guy did *not* get flamed for including both relevant, and irrelevant info and got his problem solved really quick. If one could only get coming generations to see this problem report too... -- Janne Johansson Sektionen fvr IT Media, Stockholms Universitet Frescati Hagvdg 10 106 91 STOCKHOLM http://www.it.su.se
