Remove the inconsistent .exe suffix in strace and umount usage lines.
Tidy up some trailing whitespace.
Tabs inside <screen> are not consistently formatted by all formatters, replace
with spaces.
Remove pointlesss and incorrect date
2015-06-12 Jon Turney <jon.turney@dronecode.org.uk>
* cygwin-ug-net.xml: Remove incorrect unused date.
* utils.xml : Remove .exe suffix inconsistently added in a few
places. Tidy up some trailing whitespace.
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
It seems that xidepend doesn't work correctly if we are ./configure'd using a
relative pathname to the srcdir:
$ make
cd ../../../../src/winsup/doc && ./xidepend ../../../../src/winsup/doc/cygwin-ug-net.xml ../../../../src/winsup/doc/cygwin-api.xml >"/wip/cygwin/build/x86_64-unknown-cygwin/winsup/doc/Makefile.dep"
grep: ../../../../src/winsup/doc/cygwin-ug-net.xml: No such file or directory
grep: ../../../../src/winsup/doc/cygwin-api.xml: No such file or directory
Although it might be better to fix this by making xidepend use pathnames, rather
than ignoring them and assuming everything is in the current directory...
2015-06-12 Jon Turney <jon.turney@dronecode.org.uk>
* xidepend: Fix to handle relative pathnames.
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
* fhandler_socket.cc (LOCK_EVENTS): Don't enter critical section with
invalid mutex handle since then socket has been closed.
(UNLOCK_EVENTS): Close critical section.
(fhandler_socket::evaluate_events): Handle calling connect on shutdown
socket.
(fhandler_socket::wait_for_events): Try for pthread_testcancel in case
of WAIT_FAILED. Try to come up with a better errno in case we waited
on an invalid handle.
(fhandler_socket::release_events): Change wsock_mtx and wsock_evt to
NULL under lock to avoid accessing invalid handle.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
* exceptions.cc (_cygtls::call_signal_handler): Disable enforcing
SA_RESTART in non-main threads to allow returning with EINTR from
system calls.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Not sure if this is wanted, but on a couple of occasions recently I have been
presented with strace output which contains an exception at an address in an
unknown module (i.e. not in the cygwin DLL or the main executable), so here is a
patch which adds some more information, including DLL load addresses, to help
interpret such straces.
v2:
Use NtQueryObject() for HANDLE -> filename conversion
Add new '-e' option to toggle this additional logging
2015-06-07 Jon Turney <jon.turney@dronecode.org.uk>
* strace.cc (proc_child): Log process and thread create and exit,
and DLL load and unload.
(GetFileNameFromHandle): New function.
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
* ps.cc (main): Widen UID field in long format to accommodate longer
UIDs since Cygwin 1.7.34. Remove Windows 2000 considerations. Fix
comments accordingly.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
* pinfo.cc (_pinfo::root): Fake default root for native processes.
(open_commune_proc_parms): New helper function to access process
parameter block.
(_pinfo::cwd): Fetch missing cwd for native processes from processes
parameter block.
(_pinfo::cmdline): Ditto for command line.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
This used to be used by cygwin-doc to make a tarball which would be used with a
ssh script to run docbook tools on a linux host since they weren't available on
Cygwin or something crazy like that...
2015-06-04 Jon Turney <jon.turney@dronecode.org.uk>
* Makefile.in: Remove ancient unused rules to make a documentation
tarball.
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
An estimate of the size of installing of "hundreds of megabytes" is wildly out
of date. Just remove mention of installing everything, since it is nearly
always not a good idea.
2015-06-01 Jon Turney <jon.turney@dronecode.org.uk>
* ov-ex-unix.xml: Remove unhelpful mention of and inaccurate size
estimate for installing everything.
* ov-ex-win.xml: Ditto.
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
- With the gzll kernel we have two different loading options:
- If the image is loaded to the global memory, the bootstrapping
loads the kernel to local memory. Applications are loaded on
demand. The heap then starts right after bss.
- If the image is pre-loaded to the local memory it includes the
application binaries right after bss. The heap then starts after
the application objects.
- We can check if this is a gzll kernel as it has the string "gzll" at
0x2000. At 0x200c we then find the end of the application objects in
the image. If there is no global memory we set _or1k_heap_start to
this value.
* or1k/boards/optimsoc.S: Heap for gzll kernel
- Previously the heap started right after the bss section. This can now
be configured by changing the _or1k_heap_start symbol that defaults to
the old value (&end). In board_init_early, we can now set this to
another value.
* or1k/sbrk.c: Allow for different heap start
- Store the exception program counter (from EPCR) and exception status
register (from ESR) also during the exception. A runtime system may
replace them thereby to implement a thread switch.
* or1k/exception-asm.S: Store missing state
- We do not need a red zone here, as we do not operate on the current
stack, but always use the clear exception stack. Also reserve two
extra words for the context to store EPCR and ESR.
* or1k/crt0.S: Fix exception stack frame
* or1k/exception-asm.S: ditto
- During interrupt handling the PICSR, table pointers and current
interrupt line have been saved in incorrect registers and/or stored on
the stack.
- Save the pointer in r16/r18, PICSR in r20 and the current interrupt
line in r22. Those are callee-saved registers, so that the register
values will be preserved.
* or1k/interruts-asm.S: Change registers to callee-saved.
* msp430/msp430-sim.ld (.stack): Add an assertion to make sure
that the data area does not overrun the stack. PROVIDE a new
symbol __stack_size to allow the user to set the limit.
* msp430/msp430xl-sim.ld (.stack): Likewise.
* rl78/rl78-sim.ld (.stack): Likewise.
* rl78/rl78.ld (.stack): Likewise.
* rx/rx-sim.ld (.stack): Likewise.
* rx/rx.ld (.stack): Likewise.
* msp430/msp430.ld: Delete.
* msp430/msp430F5438A-l.ld: Delete.
* msp430/msp430F5438A-s.ld: Delete.
* msp430/crt_movedata.S: Delete.
* msp430/Makefile.in (SCRIPTS): Remove msp430.ld.
(CRT_OBJS): Add crt_move_highdata.o.
* msp430/memmodel.h (START_CRT_FUNC): New macro.
(END_CRT_FUNC): New macro.
(WEAK_DEF): New macro.
* msp430/crt0.S: Use new macros.
(move_highdata): New code to initialise the .data section if it is
held in high memory.
* msp430/msp430-sim.ld (.data): Add .either.data.
(.rodata2): Move some read-only data sections here.
(.text): Add .either.text.
(.rodata): Add .either.rodata.
(.bss): Add .either.bss.
* msp430/msp430xl-sim.ld (MEMORY): Add HIROM.
(.rodata2): Move some read-only data sections here.
(.upper.data): New section. Include notes about how to initialise
it.
- With the gzll kernel we have two different loading options:
- If the image is loaded to the global memory, the bootstrapping
loads the kernel to local memory. Applications are loaded on
demand. The heap then starts right after bss.
- If the image is pre-loaded to the local memory it includes the
application binaries right after bss. The heap then starts after
the application objects.
- We can check if this is a gzll kernel as it has the string "gzll" at
0x2000. At 0x200c we then find the end of the application objects in
the image. If there is no global memory we set _or1k_heap_start to
this value.
* or1k/boards/optimsoc.S: Heap for gzll kernel
- Previously the heap started right after the bss section. This can now
be configured by changing the _or1k_heap_start symbol that defaults to
the old value (&end). In board_init_early, we can now set this to
another value.
* or1k/sbrk.c: Allow for different heap start