Implement epoll family system calls. This is a tiny wrapper
around kqueue() to implement epoll subset of functionality.
The kqueue user data are 32bit on i386 which is not enough for
epoll user data, so we keep user data in the proc emuldata.
Initial patch developed by rdivacky@ in 2007, then extended
by Yuri Victorovich @ r255672 and finished by me
in collaboration with mjg@ and jillies@.
Being exported through vdso the note.Linux section used by glibc
to determine the kernel version (this saves one uname call).
Temporarily disable the export of a note.Linux section until I figured
out how to change the kernel version in the note.Linux on the fly.
Change linux faccessat syscall definition to match actual linux one.
The AT_EACCESS and AT_SYMLINK_NOFOLLOW flags are actually implemented
within the glibc wrapper function for faccessat(). If either of these
flags are specified, then the wrapper function employs fstatat() to
determine access permissions.
Refund the proc emuldata struct for future use. For now move flags from
thread emuldata to proc emuldata as it was originally intended.
As we can have both 64 & 32 bit Linuxulator running any eventhandler
can be called twice for us. To prevent this move eventhandlers code
from linux_emul.c to the linux_common.ko module.
Introduce a new module linux_common.ko which is intended for the
following primary purposes:
1. Remove the dependency of linsysfs and linprocfs modules from linux.ko,
which will be architecture specific on amd64.
2. Incorporate into linux_common.ko general code for platforms on which
we'll support two Linuxulator modules (for both instruction set - 32 & 64 bit).
3. Move malloc(9) declaration to linux_common.ko, to enable getting memory
usage statistics properly.
Currently linux_common.ko incorporates a code from linux_mib.c and linux_util.c
and linprocfs, linsysfs and linux kernel modules depend on linux_common.ko.
Temporarily remove dtrace garbage from linux_mib.c and linux_util.c
Implement vdso - virtual dynamic shared object. Through vdso Linux
exposes functions from kernel with proper DWARF CFI information so that
it becomes easier to unwind through them.
Using vdso is a mandatory for a thread cancelation && cleanup
on a modern glibc.
To reduce code duplication introduce linux_copyout_rusage() method.
Use it in linux_wait4() system call and move linux_wait4() to the MI path.
While here add a prototype for the static bsd_to_linux_rusage().
Switch linuxulator to use the native 1:1 threads.
The reasons:
1. Get rid of the stubs/quirks with process dethreading,
process reparent when the process group leader exits and close
to this problems on wait(), waitpid(), etc.
2. Reuse our kernel code instead of writing excessive thread
managment routines in Linuxulator.
Implementation details:
1. The thread is created via kern_thr_new() in the clone() call with
the CLONE_THREAD parameter. Thus, everything else is a process.
2. The test that the process has a threads is done via P_HADTHREADS
bit p_flag of struct proc.
3. Per thread emulator state data structure is now located in the
struct thread and freed in the thread_dtor() hook.
Mandatory holdig of the p_mtx required when referencing emuldata
from the other threads.
4. PID mangling has changed. Now Linux pid is the native tid
and Linux tgid is the native pid, with the exception of the first
thread in the process where tid and pid are one and the same.
Ugliness:
In case when the Linux thread is the initial thread in the thread
group thread id is equal to the process id. Glibc depends on this
magic (assert in pthread_getattr_np.c). So for system calls that
take thread id as a parameter we should use the special method
to reference struct thread.
In preparation for switching linuxulator to the use the native 1:1
threads refactor kern_sched_rr_get_interval() and sys_sched_rr_get_interval().
Add a kern_sched_rr_get_interval() counterpart which takes a targettd
parameter to allow specify target thread directly by callee (new Linuxulator).
Linuxulator temporarily uses first thread in proc.
Move linux_sched_rr_get_interval() to the MI part.
In preparation for switching linuxulator to the use the native 1:1
threads introduce linux_exit() stub instead of sys_exit() call
(which terminates process).
In the new linuxulator exit() system call terminates the calling
thread (not a whole process).
amd64: allow base memory segment to start at address different than 0
Current code requires that the first physical memory segment starts at 0,
but this is not really needed. We only need to make sure the bootstrap code
and page tables for APs are allocated below 4GB.
This patch removes this requirement and allows booting a Dell R710 from
UEFI, where the first physical memory segment starts at 0x10000.
Sponsored by: Citrix Systems R&D
Reviewed by: jhb
Differential Revision: https://reviews.freebsd.org/D1417
Various changes to the registers displayed in DDB for x86.
- Fix segment registers to only display the low 16 bits.
- Remove unused handlers and entries for the debug registers.
- Display xcr0 (if valid) in 'show sysregs'.
- Add '0x' prefix to MSR values to match other values in 'show sysregs'.
- MFamd64: Display various MSRs in 'show sysregs'.
- Add a 'show dbregs' to display the value of debug registers.
- Dynamically size the column width for register values to properly
align columns on 64-bit platforms.
- Display %gs for i386 in 'show registers'.
Various fixes for stack unwinding in DDB on x86.
285773:
Remove some dead code from DDB's amd64 stack unwinder.
The amd64 port copied some code from i386 to fetch function arguments and
display them in backtraces. However, it was commented out and can't easily
be implemented since the function arguments are passed in
registers rather than on the stack in amd64. Remove it in preparation for
some bug fixes in this area.
285775:
Improve stack unwinding on i386 and amd64 after an IP fault.
If we can't find a symbol corresponding to the faulting instruction, assume
that the previously-executed function is a call and attempt to find the
calling function using the return address on the stack. Otherwise we end
up associating the last stack frame with the current call, which is
incorrect and causes the unwinder to skip printing of the calling function,
resulting in a confusing backtrace.
285776:
Let the unwinder handle faults during function prologues or epilogues.
The i386 and amd64 DDB stack unwinders contain code to detect and handle
the case where the first frame is not completely set up or torn down. This
code was accidentally unused however, since db_backtrace() was never called
with a non-NULL trap frame. This change fixes that.
Also remove get_rsp() from the amd64 code. It appears to have come from
i386, which needs to take into account whether the exception triggered a
CPL switch, since SS:ESP is only pushed onto the stack if so. On amd64,
SS:RSP is pushed regardless, so get_rsp() was doing the wrong thing for
kernel-mode exceptions. As a result, we can also remove custom print
functions for these registers.
Fix integer truncation bug in malloc(9)
A couple of internal functions used by malloc(9) and uma truncated
a size_t down to an int. This could cause any number of issues
(e.g. indefinite sleeps, memory corruption) if any kernel
subsystem tried to allocate 2GB or more through malloc. zfs would
attempt such an allocation when run on a system with 2TB or more
of RAM.
- Improve support for Macs that have a stride not equal to the
horizonal resolution (width).
- Support frame buffers that are larger than the default screen
size.
- Support large frame buffers: add 24 more page table pages we
allocate on boot-up.
PR: 193745
