Re: make: don't know how to make /usr/lib/crt0.o (prerequisite of: httpd)

[Kent Watsen]

> Does /usr/lib/crt0.o exist? If it doesn't did tar complete successfully?

I’ve since scrubbed that install, but I did check before for the file and noted 
that it was not present.  Interestingly, the file is in the TGZ, so something 
happened…

$ tar -tzvf comp70.tgz | grep /usr/lib/crt0.o 
-r--r--r--  0 root   bin  2544 Sep 30 16:00 ./usr/lib/crt0.o


> Maybe try:
> 
> # tar -C / -xzphf comp70.tgz || echo "somethings broken"
> 
> You may have run out of disk space or something similar and just didn't 
> notice. Thats my best guess.

It was a 4GB partition, so unlikely an out of disk issue.  That said, there’s a 
chance that the extraction terminated prematurely for other reasons, which your 
suggestion would help catch - thanks for the suggestion!



Kent

Re: make: don't know how to make /usr/lib/crt0.o (prerequisite of: httpd)

[Kent Watsen]

> As you're the one in possession of the system with the nonstandard
> configuration you're the best person to figure out what's different
> between that and a normally installed system.

My bad, I thought it was supported to install filesets this way, but I don’t 
see this approach discussed on openbsd.org  now.


> It does seem an unusual requirement to have a system which cannot have
> comp sets, but that is still ok to fetch a file from the internet and
> untar it as root without verifying the signify(1) signature and hash.

Point.  I already have SHA256 in the directory, so easy to add.


> Had you considered building the binary on a non production system
> instead?

I did, but since it worked with 6.9 before, figured I’d try that route first.


Thanks,
Kent

Re: make: don't know how to make /usr/lib/crt0.o (prerequisite of: httpd)

[Edgar Pettijohn]

On 10/31/21 2:23 PM, Kent Watsen wrote:

The “httpd-plus” [1] patch installs just find when a fresh 7.0 install selects packages "base", 
"bsd", "bsd.rd", "bsd.mp", “comp”, and “man”.

However, when a fresh 7.0 install selects all the same packages except “comp”, 
and then subsequently adds the “comp” package via the command:

(cd /root && curl -s -O https://cdn.openbsd.org/pub/OpenBSD/7.0/amd64/comp70.tgz 
&& cd / && tar xzvphf /root/comp70.tgz)

The installation of the "httpd-plus" patch fails with the following snippet:


Building and installing httpd-plus binary and manpage ...
    /usr/src/usr.sbin/httpd/obj -> /usr/obj/usr.sbin/httpd
make: don't know how to make /usr/lib/crt0.o (prerequisite of: httpd)


Does /usr/lib/crt0.o exist? If it doesn't did tar complete successfully? 
Maybe try:


# tar -C / -xzphf comp70.tgz || echo "somethings broken"

You may have run out of disk space or something similar and just didn't 
notice. Thats my best guess.



Edgar


Stop in /usr/src/usr.sbin/httpd
Restoring original sources ... Done.
Installing httpd-plus failed (exitcode: 2).

This logic worked on 6.9, what’s the difference?  Why can’t /usr/lib/crt0.o be 
found or made?  How to get past this error without needing to install the 
“comp” package during installation?

PS: I don’t want “comp” on the production system. After installing “httpd-plus”, I run the 
following command to remove it: (cd /root && for i in `tar -tzvf /root/comp70.tgz | awk 
'{print $NF}'`; do rm -rf $i; done) && rm /root/comp70.tgz

[1] https://github.com/mpfr/httpd-plus/tree/7.0-stable

Thanks,
Kent

Re: make: don't know how to make /usr/lib/crt0.o (prerequisite of: httpd)

[Stuart Henderson]

On 2021-10-31, Kent Watsen  wrote:
> Thanks Theo. 
>
> No debate about needing comp, only how it's installed…or maybe I 
> misunderstand what you mean by “the script”?

As you're the one in possession of the system with the nonstandard
configuration you're the best person to figure out what's different
between that and a normally installed system.

>>> However, when a fresh 7.0 install selects all the same packages except 
>>> “comp”, and then subsequently adds the “comp” package via the command:
>>> 
>>>(cd /root && curl -s -O 
>>> https://cdn.openbsd.org/pub/OpenBSD/7.0/amd64/comp70.tgz && cd / && tar 
>>> xzvphf /root/comp70.tgz)

>>> PS: I don’t want “comp” on the production system. After installing 
>>> “httpd-plus”, I run the following command to remove it: (cd /root && for i 
>>> in `tar -tzvf /root/comp70.tgz | awk '{print $NF}'`; do rm -rf $i; done) && 
>>> rm /root/comp70.tgz

It does seem an unusual requirement to have a system which cannot have
comp sets, but that is still ok to fetch a file from the internet and
untar it as root without verifying the signify(1) signature and hash.

Had you considered building the binary on a non production system
instead?


-- 
Please keep replies on the mailing list.

Re: make: don't know how to make /usr/lib/crt0.o (prerequisite of: httpd)

[Kent Watsen]

Thanks Theo. 

No debate about needing comp, only how it's installed…or maybe I misunderstand 
what you mean by “the script”?

Cheers,
Kent


> On Oct 31, 2021, at 3:38 PM, Theo de Raadt  wrote:
> 
> From the script
> 
> make obj && make && make install
> 
> Which uses the whole toolchain.
> 
> You need comp.  You don't have a choice.
> 
> Kent Watsen  wrote:
> 
>> The “httpd-plus” [1] patch installs just find when a fresh 7.0 install 
>> selects packages "base", "bsd", "bsd.rd", "bsd.mp", “comp”, and “man”.
>> 
>> However, when a fresh 7.0 install selects all the same packages except 
>> “comp”, and then subsequently adds the “comp” package via the command:
>> 
>>(cd /root && curl -s -O 
>> https://cdn.openbsd.org/pub/OpenBSD/7.0/amd64/comp70.tgz && cd / && tar 
>> xzvphf /root/comp70.tgz)
>> 
>> The installation of the "httpd-plus" patch fails with the following snippet:
>> 
>>   
>>Building and installing httpd-plus binary and manpage ...
>>/usr/src/usr.sbin/httpd/obj -> /usr/obj/usr.sbin/httpd
>>make: don't know how to make /usr/lib/crt0.o (prerequisite of: httpd)
>>Stop in /usr/src/usr.sbin/httpd
>>Restoring original sources ... Done.
>>Installing httpd-plus failed (exitcode: 2).
>> 
>> This logic worked on 6.9, what’s the difference?  Why can’t /usr/lib/crt0.o 
>> be found or made?  How to get past this error without needing to install the 
>> “comp” package during installation?
>> 
>> PS: I don’t want “comp” on the production system. After installing 
>> “httpd-plus”, I run the following command to remove it: (cd /root && for i 
>> in `tar -tzvf /root/comp70.tgz | awk '{print $NF}'`; do rm -rf $i; done) && 
>> rm /root/comp70.tgz
>> 
>> [1] https://github.com/mpfr/httpd-plus/tree/7.0-stable
>> 
>> Thanks,
>> Kent
>> 
>> 

Re: make: don't know how to make /usr/lib/crt0.o (prerequisite of: httpd)

[Theo de Raadt]

>From the script

 make obj && make && make install

Which uses the whole toolchain.

You need comp.  You don't have a choice.

Kent Watsen  wrote:

> The “httpd-plus” [1] patch installs just find when a fresh 7.0 install 
> selects packages "base", "bsd", "bsd.rd", "bsd.mp", “comp”, and “man”.
> 
> However, when a fresh 7.0 install selects all the same packages except 
> “comp”, and then subsequently adds the “comp” package via the command:
> 
>   (cd /root && curl -s -O 
> https://cdn.openbsd.org/pub/OpenBSD/7.0/amd64/comp70.tgz && cd / && tar 
> xzvphf /root/comp70.tgz)
> 
> The installation of the "httpd-plus" patch fails with the following snippet:
> 
>
>   Building and installing httpd-plus binary and manpage ...
>   /usr/src/usr.sbin/httpd/obj -> /usr/obj/usr.sbin/httpd
>   make: don't know how to make /usr/lib/crt0.o (prerequisite of: httpd)
>   Stop in /usr/src/usr.sbin/httpd
>   Restoring original sources ... Done.
>   Installing httpd-plus failed (exitcode: 2).
> 
> This logic worked on 6.9, what’s the difference?  Why can’t /usr/lib/crt0.o 
> be found or made?  How to get past this error without needing to install the 
> “comp” package during installation?
> 
> PS: I don’t want “comp” on the production system. After installing 
> “httpd-plus”, I run the following command to remove it: (cd /root && for i in 
> `tar -tzvf /root/comp70.tgz | awk '{print $NF}'`; do rm -rf $i; done) && rm 
> /root/comp70.tgz
> 
> [1] https://github.com/mpfr/httpd-plus/tree/7.0-stable
> 
> Thanks,
> Kent
> 
> 

make: don't know how to make /usr/lib/crt0.o (prerequisite of: httpd)

[Kent Watsen]

The “httpd-plus” [1] patch installs just find when a fresh 7.0 install selects 
packages "base", "bsd", "bsd.rd", "bsd.mp", “comp”, and “man”.

However, when a fresh 7.0 install selects all the same packages except “comp”, 
and then subsequently adds the “comp” package via the command:

(cd /root && curl -s -O 
https://cdn.openbsd.org/pub/OpenBSD/7.0/amd64/comp70.tgz && cd / && tar xzvphf 
/root/comp70.tgz)

The installation of the "httpd-plus" patch fails with the following snippet:

   
Building and installing httpd-plus binary and manpage ...
    /usr/src/usr.sbin/httpd/obj -> /usr/obj/usr.sbin/httpd
make: don't know how to make /usr/lib/crt0.o (prerequisite of: httpd)
Stop in /usr/src/usr.sbin/httpd
Restoring original sources ... Done.
Installing httpd-plus failed (exitcode: 2).

This logic worked on 6.9, what’s the difference?  Why can’t /usr/lib/crt0.o be 
found or made?  How to get past this error without needing to install the 
“comp” package during installation?

PS: I don’t want “comp” on the production system. After installing 
“httpd-plus”, I run the following command to remove it: (cd /root && for i in 
`tar -tzvf /root/comp70.tgz | awk '{print $NF}'`; do rm -rf $i; done) && rm 
/root/comp70.tgz

[1] https://github.com/mpfr/httpd-plus/tree/7.0-stable

Thanks,
Kent

Re: Make don't know how to make

[Alex Carver]

Matthias Kilian wrote:

On Thu, Mar 18, 2010 at 07:14:36PM -0700, Alexander Carver wrote:
The error isn't always the same file on two consecutive tries but they do 
seem to repeat themselves (in other words, init_sysent.c has shown up as 
an error more than once but not consecutively, same for pf.c)

Intermittent and inconsistent signal 11's from the compiler: that's
classically the sign of bad memory in the box.  Try removing or
replacing the memory.

Ok, I've gone through multiple iterations of memory sticks of various 
sizes and even swapped motherboards but I still get signal 11's.  Is 
there a low memory flag I can set that will prevent make from using all 
available RAM?  This signal 11 issue doesn't crop up immediately, it 
takes quite some time before it finally shows up.


Did you also try changing the power supply? Even if spurious
segmentation faults are *often* caused by bad memory, bad mainboards,
overheated cpus etc., they *may* also be caused by a bad power
supply.


Yes, I did change the power supply, too.  I have two identical machines 
so I swapped the RAM and hard drive over from one to the other.  I also 
have both machines plugged into an AC line filter to clean up the main 
power.

Re: Make don't know how to make

[Alexander Carver]

Philip Guenther wrote:

On Friday, March 5, 2010, Alex Carver  wrote:
...

Assembler messages:
Warning: end of file not at end of a line; newline inserted
cpp0: output pipe has been closed
cc: Internal compiler error: program cc1 got fatal signal 11
[standard input]:2197: Error: Illegal operands

The error isn't always the same file on two consecutive tries but they do seem 
to repeat themselves (in other words, init_sysent.c has shown up as an error 
more than once but not consecutively, same for pf.c)


Intermittent and inconsistent signal 11's from the compiler: that's
classically the sign of bad memory in the box.  Try removing or
replacing the memory.



Ok, I've gone through multiple iterations of memory sticks of various 
sizes and even swapped motherboards but I still get signal 11's.  Is 
there a low memory flag I can set that will prevent make from using all 
available RAM?  This signal 11 issue doesn't crop up immediately, it 
takes quite some time before it finally shows up.

Re: Make don't know how to make

[Nick Holland]

Alexander Carver wrote:
 Philip Guenther wrote:
 On Friday, March 5, 2010, Alex Carver  wrote:
 ...
 Assembler messages:
 Warning: end of file not at end of a line; newline inserted
 cpp0: output pipe has been closed
 cc: Internal compiler error: program cc1 got fatal signal 11
 [standard input]:2197: Error: Illegal operands

 The error isn't always the same file on two consecutive tries but they do 
 seem to repeat themselves (in other words, init_sysent.c has shown up as an 
 error more than once but not consecutively, same for pf.c)
 
 Intermittent and inconsistent signal 11's from the compiler: that's
 classically the sign of bad memory in the box.  Try removing or
 replacing the memory.
 
 
 Ok, I've gone through multiple iterations of memory sticks of various 
 sizes and even swapped motherboards but I still get signal 11's.  Is 
 there a low memory flag I can set that will prevent make from using all 
 available RAM?  This signal 11 issue doesn't crop up immediately, it 
 takes quite some time before it finally shows up.

some machines just do this.
I've had horrible problems with this on several AMD K6 systems, the
solution seems to be (on those, if possible) underclock the external
bus, and increase the clock multiplier.  Took a 450MHz system
(4.5x100MHz probably), changed it to 5.5x83.3MHz), and the thing has
been rock solid ever since.  Actually, it was rock solid before, for
everything but compiling.

I'm guessing some of the glue chips between the CPU and the RAM
suck, at least the way GCC beats on 'em.  It wasn't the RAM itself,
swapped with good in other systems RAM, even with 133MHz parts.

Nick.

Re: Make don't know how to make

[Alexander Carver]

Nick Holland wrote:

Alexander Carver wrote:


Ok, I've gone through multiple iterations of memory sticks of various 
sizes and even swapped motherboards but I still get signal 11's.  Is 
there a low memory flag I can set that will prevent make from using all 
available RAM?  This signal 11 issue doesn't crop up immediately, it 
takes quite some time before it finally shows up.


some machines just do this.
I've had horrible problems with this on several AMD K6 systems, the
solution seems to be (on those, if possible) underclock the external
bus, and increase the clock multiplier.  Took a 450MHz system
(4.5x100MHz probably), changed it to 5.5x83.3MHz), and the thing has
been rock solid ever since.  Actually, it was rock solid before, for
everything but compiling.

I'm guessing some of the glue chips between the CPU and the RAM
suck, at least the way GCC beats on 'em.  It wasn't the RAM itself,
swapped with good in other systems RAM, even with 133MHz parts.

Nick.


Unfortunately on this little IPX I can't change any speeds.  I was 
hoping for some flag that could run the compiler in low memory mode.


I suppose I could just remove the output file that failed and resume the 
make.  It seems to go through again just fine and then sometime later 
messes up.  One way or another I need to get this kernel compiled so I 
can try the supplied patches for /dev/cua* devices.

Re: Make don't know how to make

[Matthias Kilian]

On Thu, Mar 18, 2010 at 07:14:36PM -0700, Alexander Carver wrote:
 The error isn't always the same file on two consecutive tries but they do 
 seem to repeat themselves (in other words, init_sysent.c has shown up as 
 an error more than once but not consecutively, same for pf.c)
 
 Intermittent and inconsistent signal 11's from the compiler: that's
 classically the sign of bad memory in the box.  Try removing or
 replacing the memory.
 
 
 Ok, I've gone through multiple iterations of memory sticks of various 
 sizes and even swapped motherboards but I still get signal 11's.  Is 
 there a low memory flag I can set that will prevent make from using all 
 available RAM?  This signal 11 issue doesn't crop up immediately, it 
 takes quite some time before it finally shows up.

Did you also try changing the power supply? Even if spurious
segmentation faults are *often* caused by bad memory, bad mainboards,
overheated cpus etc., they *may* also be caused by a bad power
supply.

Ciao,
Kili

Re: Make don't know how to make

[Alexander Carver]

Philip Guenther wrote:

On Friday, March 5, 2010, Alex Carver  wrote:
...

Assembler messages:
Warning: end of file not at end of a line; newline inserted
cpp0: output pipe has been closed
cc: Internal compiler error: program cc1 got fatal signal 11
[standard input]:2197: Error: Illegal operands

The error isn't always the same file on two consecutive tries but they do seem 
to repeat themselves (in other words, init_sysent.c has shown up as an error 
more than once but not consecutively, same for pf.c)


Intermittent and inconsistent signal 11's from the compiler: that's
classically the sign of bad memory in the box.  Try removing or
replacing the memory.


I'll try that again although I did it once before with no luck.  But 
it's worth another shot.

Re: Make don't know how to make

[Vadim Zhukov]

On 5 March 2010 c. 05:52:49 Alexander Carver wrote:
 Abel Abraham Camarillo Ojeda wrote:
  Alex Carver agcarver+open...@acarver.net wrote:
  Suggestions?
 
  are you running current? if so:
 
  http://www.openbsd.org/faq/current.html#20100203

 No, I am not running current.  This is a stock installation of 4.6 and
 the source is from the src.tar.gz file from the same place.

 I went ahead and wiped the entire src tree and started from scratch.
 The compile started working but then died later:

 In file included from ../../../../kern/init_sysent.c:14:
 ../../../../sys/syscallargs.h:1119: Internal compiler error in
 'grokdeclarator', at c-decl.c:4505

 No idea what happened there.

1. Please show your /etc/mk.conf .

2. Does this error always occur at the same place?

--
  Best wishes,
Vadim Zhukov

A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?

Re: Make don't know how to make

[Alex Carver]

Vadim Zhukov wrote:

On 5 March 2010 c. 05:52:49 Alexander Carver wrote:

Abel Abraham Camarillo Ojeda wrote:

Alex Carver agcarver+open...@acarver.net wrote:

Suggestions?

are you running current? if so:

http://www.openbsd.org/faq/current.html#20100203

No, I am not running current.  This is a stock installation of 4.6 and
the source is from the src.tar.gz file from the same place.

I went ahead and wiped the entire src tree and started from scratch.
The compile started working but then died later:

In file included from ../../../../kern/init_sysent.c:14:
../../../../sys/syscallargs.h:1119: Internal compiler error in
'grokdeclarator', at c-decl.c:4505

No idea what happened there.


1. Please show your /etc/mk.conf .


I don't have an mk.conf file anywhere on the system.  It is a completely 
fresh installation.




2. Does this error always occur at the same place?


The error doesn't seem to happen at the same place every time but it has 
happened there before.  I just cleared the tree and started again.  This 
time  I received the error when processing ../../../../net/pf.c:


Assembler messages:
Warning: end of file not at end of a line; newline inserted
cpp0: output pipe has been closed
cc: Internal compiler error: program cc1 got fatal signal 11
[standard input]:2197: Error: Illegal operands

The error isn't always the same file on two consecutive tries but they 
do seem to repeat themselves (in other words, init_sysent.c has shown up 
as an error more than once but not consecutively, same for pf.c)

Re: Make don't know how to make

[Philip Guenther]

On Friday, March 5, 2010, Alex Carver  wrote:
...
 Assembler messages:
 Warning: end of file not at end of a line; newline inserted
 cpp0: output pipe has been closed
 cc: Internal compiler error: program cc1 got fatal signal 11
 [standard input]:2197: Error: Illegal operands

 The error isn't always the same file on two consecutive tries but they do 
 seem to repeat themselves (in other words, init_sysent.c has shown up as an 
 error more than once but not consecutively, same for pf.c)

Intermittent and inconsistent signal 11's from the compiler: that's
classically the sign of bad memory in the box.  Try removing or
replacing the memory.


Philip Guenther

Re: Make don't know how to make

[Alexander Carver]

Abel Abraham Camarillo Ojeda wrote:

Alex Carver agcarver+open...@acarver.net wrote:


Suggestions?


are you running current? if so:

http://www.openbsd.org/faq/current.html#20100203


No, I am not running current.  This is a stock installation of 4.6 and 
the source is from the src.tar.gz file from the same place.


I went ahead and wiped the entire src tree and started from scratch. 
The compile started working but then died later:


In file included from ../../../../kern/init_sysent.c:14:
../../../../sys/syscallargs.h:1119: Internal compiler error in 
'grokdeclarator', at c-decl.c:4505


No idea what happened there.

Make don't know how to make

[Alex Carver]

I'm trying to recompile my generic sparc kernel to make use of some 
/dev/cua* patches that were sent to me on the list not too long ago but 
now I'm running into a problem with make.


When I first received the patches, things were mostly working but I was 
running out of memory and swap space (small machine) so a lot of the 
build was failing.  But make did work at that time.


Now, the only thing different about the system is that I added another 
hard drive and mounted the entire space as swap.


The root is on disk /dev/sd1a.  That drive is a bit small so there's a 
second drive that is holding the source code for the kernel.  It is 
/dev/sd2a and is mounted at /disk2.  I also have a symlink from 
/usr/src/sys pointing to /disk2/sys where the kernel source is located.


When I first tried (prior to discovering I ran out of memory and swap), 
I was going to /usr/src/sys/arch/sparc/compile/GENERIC (I had already 
run config GENERAL) and then make clean  make depend  make. 
Everything seemed to work ok other than running out of memory.


Now, after the new swap drive (/dev/sd3a) was added, I get the following:

make clean works

make depend:
make: don't know how to make ../../../../arch/sparc/sparc/auxreg.c. Stop 
in /disk2/sys/arch/sparc/compile/GENERIC.


Suggestions?

Re: Make don't know how to make

[Abel Abraham Camarillo Ojeda]

Alex Carver agcarver+open...@acarver.net wrote:

 I'm trying to recompile my generic sparc kernel to make use of some 
 /dev/cua* patches that were sent to me on the list not too long ago but 
 now I'm running into a problem with make.

 When I first received the patches, things were mostly working but I was 
 running out of memory and swap space (small machine) so a lot of the 
 build was failing.  But make did work at that time.

 Now, the only thing different about the system is that I added another 
 hard drive and mounted the entire space as swap.

 The root is on disk /dev/sd1a.  That drive is a bit small so there's a 
 second drive that is holding the source code for the kernel.  It is 
 /dev/sd2a and is mounted at /disk2.  I also have a symlink from 
 /usr/src/sys pointing to /disk2/sys where the kernel source is located.

 When I first tried (prior to discovering I ran out of memory and swap), 
 I was going to /usr/src/sys/arch/sparc/compile/GENERIC (I had already 
 run config GENERAL) and then make clean  make depend  make. 
 Everything seemed to work ok other than running out of memory.

 Now, after the new swap drive (/dev/sd3a) was added, I get the following:

 make clean works

 make depend:
 make: don't know how to make ../../../../arch/sparc/sparc/auxreg.c. Stop 
 in /disk2/sys/arch/sparc/compile/GENERIC.

 Suggestions?

are you running current? if so:

http://www.openbsd.org/faq/current.html#20100203

--
DISCLAIMER: http://goldmark.org/jeff/stupid-disclaimers/ 
This message will self-destruct in 3 seconds.

Re: Make don't know how to make

[Miod Vallat]

 make depend:
 make: don't know how to make ../../../../arch/sparc/sparc/auxreg.c. Stop in 
 /disk2/sys/arch/sparc/compile/GENERIC.

 Suggestions?

Try `rm -f .depend' before running `make depend'.