Re: devel/llvm13 failed to reclaim memory on 8 GB Pi4 running -current [UFS context: used the whole swap space too]

On 2022-Jan-28, at 17:14, bob prohaska <fbsd@www.zefox.net> wrote:

> On Fri, Jan 28, 2022 at 04:33:11PM -0800, Mark Millard wrote:
>> 
>> 
>> The part of it that has:
>> 
>> CFLAGS.clang+=  -O0
>> .if empty(DEBUG_FLAGS)
>> MK_DEBUG_FILES:=no
>> CFLAGS.clang+=  -g0
>> .endif
>> 
>> is important to limiting memory use for building
>> googletest.
>> 
> 
> The version on my Pi3 running stable/13 contains
> CFLAGS.clang+=  -O0
> .if empty(DEBUG_FLAGS)
> MK_DEBUG_FILES:=no
> CFLAGS.clang+=  -g0
> .endif
> 
> No differences that I can see.

Okay.

I'll note that flang is new to devel/llvm13 and,
for comparison to historical build-clang memory
use, I ignore flang. Building flag is not
involved in buildworld. But the failures that you
have reported were not building clang or anything
from the llvm project. They were building gmock
things.



An interesting thing that I've noted in your reporting:

c++: note: diagnostic msg: /tmp/gmock-all-836ef8.cpp
c++: note: diagnostic msg: /tmp/gmock-all-836ef8.sh

vs. http://www.zefox.net/~fbsd/rpi3/20220121/ which has:

	• gmock_main-f5c28a.cpp
	• gmock_main-f5c28a.sh

Those are different .cpp files and I only have access
to the gmock_main-f5c28a.cpp related pair of files.
I've never gotten a failure for what I have access to.


You have not reported the lldb result for my reply that
listed doing:

QUOTE
To look at the different frames:

(lldb) up 1
(lldb) bt
. . . (output) . . .
(lldb) up 1
(lldb) bt
. . . (output) . . .

and so on until #5 has been displayed.
END QUOTE

This was a gmock_main-f5c28a.cpp context. The "up"
commands make the sequence more than just "repeated
bt commands". It should show the source code of the
lines of code in question for the specific frame,
much like it did for frame #0 in what you had sent
to the list.
 

===
Mark Millard
marklmi at yahoo.com