Results in: uptime: couldn't get boot time: Not enough space

Typing 'logout' or ctrl-d keeps the connection alive, you can still type, but nothing happens.

[03:05:27] ~ $ /include/alloca.h
/include/alloca.h: line 1: /applications: is a directory
/include/alloca.h: line 3: /applications: is a directory

And then the system freezes.

If you 'cd /dev', you end up on 'dev»/', and suddenly can't access any of the files on 'root»/' without a prefix. Because PATH is usually '/applications', not 'root»/applications', it suddenly becomes very difficult to run any programs.

Ideally getcwd() should also include the mount (eg, 'root»/') to make this more obvious.

If you try to execute a shell script (such as a configure script), or even a non shell script such as a C code file, then the DynamicLinker will try to execute the file and crash and burn as you would expect.

When executing a code file: breakpoint as it knows it aint an ELF.
When executing a shell script: pagefault

Running seq with no arguments causes a segfault, running with a numeric argument results in no output.

I found the code associated with IOAPIC not implemented.
How to assign an interrupt to the multiprocessor in 64-bit architecture?

sleep returns immediately, without sleeping.

Running stty with no arguments, or 'stty onlcr' causes segmentation fault.

I ran 'yes' on one connection, and was then unable to type into the session (ctrl-c didn't work), but I was able to log in on a new ssh connection.

Something for pedigree-apps

Running 'env' results in a segmentation fault.

Running gcc with no arguments (or with) causes segfault.

Running /libraries/libc.so causes system to freeze (over ssh).

Build line:

scons

Causes output:

Linking build/src/system/kernel/kernel
/home/james/.pedigree_2/bin/i686-elf-ld: -f may not be used without -shared
scons: *** [build/src/system/kernel/kernel] Error 1
scons: building terminated because of errors.

TimeoutGuard's destructor calls Thread::cullEvent if the event did not fire when it goes out of scope.

However, cullEvent currently has a FATAL("cullEvent"); line in it, making it impossible to use TimeoutGuard properly. I'm not sure why it has this line, but I assume there's a good reason.

Failing code:
Semaphore sem(0);

{
    TimeoutGuard guard(5);
    if (!guard.timedOut())
        NOTICE("Acquired sem");
}

NOTICE("Ended!");
for(;;);

If the semaphore acquire is not commented the timeout occurs as per normal, and there is no breakpoint.

x86_64 GCC page faults on develop branch.

I modified the script file:

!/bin/bash

SERIAL_PORTS="-serial file:./../build/serial.txt -serial file:./../build/memlog.txt"

if [ ! -z "$NO_SERIAL_PORTS" ]; then
SERIAL_PORTS=""
fi

AIO="native"

if [ ! "$(uname -s)" == "Linux" ]; then
AIO="threads"
fi

qemu-system-x86_64 -smp 4 -m 512
-drive file=./../build/hdd.img,if=ide,media=disk,cache=writeback,aio=$AIO,index=0,format=raw
-drive file=./../build/pedigree.iso,if=ide,media=cdrom,cache=writeback,aio=$AIO,index=2,format=raw
-boot d -vga std $SERIAL_PORTS -rtc base=utc $*

However, the system enters the exception handler...

The build scripts rely on http://pedigree-project.org/ existing and providing extra libraries required to build this code. Since this site is down, the code can't be built.

May I suggest that all the required dependencies be provided on GitHub?