XXX: When CLONE_VM is set but CLONE_THREAD is not the new thread is
meant to have its own thread group, i.e., when calling exit_group()
the cloner thread wouldn't be killed. However, this is a problem on
the Linux side because we do not invoke clone in mcexec when threads
are created. Thus, currently no support for this combination is
provided.
required so that a forked process can obtain exec reference in the
Linux kernel even if executable was specified with relative path
and fork was called after changing the current working directory
It decides the number of cores for McKernel by looking into the
"SHIMOS: CPU Status:" line of dmesg. It sets the amount of memory for
McKernel to one urth of the total memory obtained by "free -g".
There is a problem which causes wait4(2) hang when wait4(2) called by a
process races with its child process termination.
This is a quick fix for this problem. See idle().
refs #271
Request pages, response pages, and a doorbell page are mapped into
process space to enable user processes to delegate system calls
directly to mcctrl/mcexec.
This commit removes these mappings for the following reasons.
- These mappings cause a memory leak in current fork() implementation.
- These mappings are not used.
- These mappings do not function properly.
And the fix which corrects function of these mappings is not easy.
We didn't have any lock variable to guard the manipulation of a PTE for
kernel virual address space. Note that we have one for user virtual
address space.
* support "memdebug" mckernel option
* check buffer overrun when memory free and next mcexec run
* check double free
* check memory leak when next mcexec run
The page-fault handler called on the Linux side looks for a matching PTE
in the page table residing in McKernel. If it fails to find it, it
wrote a confusing kernel message. It's supressed now because it's not an
error.
