again. This was changed in sys/socket.h r1.51 to work around fallout
from the IPv6 aux data migration. It broke the historic ABI on some
platforms. This commit restores compatibility for netbsd32 code on such
platforms and provides a template for future changes to the CMSG_*
alignment. Revert PCC/Clang workarounds in postfix and tmux.
changes to the atf codebase with upstream code. All local changes should
have been backported, with appropriate portability workarounds where
necessary.
This new release also includes other changes though, such as performance
improvements and bug fixes, and also a major new feature partially
implemented by pgoyette@: the time to execute each test cases and test
program is now recorded and included in the output reports.
The import into NetBSD has been tested natively on amd64 and macppc, and
the full test suite has also been run through anita on amd64 and i386.
No regressions observed... but you never know.
From the NEWS file, the changes in this version are as follows:
Experimental version released on January 16th, 2012.
* Respect stdin in atf-check. The previous release silenced stdin for any
processes spawned by atf, not only test programs, which caused breakage
in tests that pipe data through atf-check.
* Performance improvements to atf-sh.
* Enabled detection of unused parameters and variables in the code and
fixed all warnings.
* Changed the behavior of "developer mode". Compiler warnings are now
enabled unconditionally regardless of whether we are in developer mode or
not; developer mode is now only used to perform strict warning checks and
to enable assertions. Additionally, developer mode is now only
automatically enabled when building from the repository, not for formal
releases.
* Added new Autoconf M4 macros (ATF_ARG_WITH, ATF_CHECK_C and
ATF_CHECK_CXX) to provide a consistent way of defining a --with-arg flag
in configure scripts and detecting the presence of any of the ATF
bindings. Note that ATF_CHECK_SH was already introduced in 0.14, but it
has now been modified to also honor --with-atf if instantiated.
* Added timing support to atf-run / atf-report.
* Added support for a 'require.memory' property, to specify the minimum
amount of physical memory needed by the test case to yield valid results.
* PR bin/45690: Force an ISO-8859-1 encoding in the XML files generated by
atf-report so that invalid data in the output of test cases does not
mangle our report.
The full changelog is at
http://pcc.ludd.ltu.se/fisheye/changelog/~date=2012-01-05T06%3A00%3A00/pcc
notable improvements since 2011-09-01
- the previous version failed to compile an i386 kernel, this works
(and the kernel boots though I would consider it experimental)
- MacOS X fixes
- misc documentation improvements
- improved regression tests
- some work has been done towards supporting C++
GCC can generate bogus dwarf attributes with DW_AT_byte_size set to 0xFFFFFFFF.
See http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35998 .
GCC is currently doing this for external/bsd/tmux/dist/compat/imsg-buffer.c:
readelf -a --debug-dump imsg-buffer.o
...
<2><6e3>: Abbrev Number: 32 (DW_TAG_union_type)
<6e4> DW_AT_byte_size : 0xffffffff
<6e8> DW_AT_decl_file : 1
<6e9> DW_AT_decl_line : 229
<6ea> DW_AT_sibling : <0x705>
This resulted in ctfconvert generating a faulty CTF entry which then caused the
segfault in ctfmerge.
The fix has ctfconvert check for the bogus 0xFFFFFFFF value and works around it.
It also adds some protection to ctfmerge to avoid the segfault and fail
more gracefully if the error should occur in the future.
- fix the handling of subdir additional LIBDPLIBS so they're present
when <bsd.lib.mk> is loaded. fixes missing shlib links relative to
what the normal package does. (noted by wiz.)
Also fix a typo/pasto when checking devices against devices/types in the
configuration file: use value from configuration, not the end of the
device_info table. Devices which aren't explicitly specified in the
device_info table can now be used by adding a type entry in the config
file.
ENHANCE: removed static build autodetection, which didn't work anyway.
Use an explicit, user-specified preprocessor variable instead.
ENHANCE: cleaned up the documentation a bit.
ENHANCE: added openpam_subst(3), allowing certain PAM items to be embedded
in strings such as prompts. Apply it to the prompts used by
pam_get_user(3) and pam_get_authtok(3).
ENHANCE: added support for the user_prompt, authtok_prompt and
oldauthtok_prompt module options, which override the prompts passed
by the module to pam_set_user(3) and pam_get_authtok(3).
ENHANCE: rewrote the policy parser to support quoted option values.
ENHANCE: added pamtest(1), a tool for testing modules and policies.
ENHANCE: added code to check the ownership and permissions of a module before
loading it.
ENHANCE: added / improved input validation in many cases, including the policy
file and some function arguments.
1. include <net/dlt.h> where needed since we have the information in a separate
file.
2. #error in pcap/bpf.h to make sure it is not being used.
3. If we don't include <pcap/bpf.h> include <net/bpf.h> instead.
4. Install symlinks instead of copies of files.
and joerg@'s enable tls change.
missing: m68010 (coming shortly) and m68k (breaks the build)
XXX: didn't test i re-merged the powerpc by-hand changes properly
in the NetBSD sources. This should fix build problems for people using
GIT to track current.
Problem reported by Evgeniy Ivanov on "current-users" mailing list.
built without TLS support, at least not without major surgery.
I've only tested this by building with "MKCRYPTO" set to "yes"
because the build fails much ealier otherwise.
Problem reported by Nick Hudson in private e-mail.
that we can find their version from verdef not vertree. This bug
showed up on the hp700 build of heimdal's libkrb5, where we got
the "end" symbol with a bad version from libhx509. The hp700 linker
script defines that end symbol.
* More printf portabitiy fixes.
* Use read -r to avoid backslash problems.
* If we have a valid domain, put that in resolv.conf as well as search.
This does not fix a technical problem, just stops me getting bug reports.
* Update metric and privacy even if resolv.conf didn't change.
* sortlist is now supported.
* Ensure subscriber config directories exist before writing the configs
* Don't create pdnsd.conf if it doesn't exist or is not writeable.
Postfix stable release 2.8.7 is available. This contains a workaround for
a problem that is fixed in Postfix 2.9.
- The postscreen daemon, which is not enabled by default, sent non-compliant
SMTP responses (220- followed by 421) when it could not give a connection
to a real smtpd process. These responses caused some remote SMTP clients
to return mail as undeliverable.
The workaround is to hang up after sending 220- without sending the
421 "sorry" reply; this is harmless.
Don't abort for now. This worked before because the previous version of gcc
did not bother unregistering. Adding debugging printfs makes emacs core
dump, where other programs work fine.
- split alloca stuff in a separate file
- fix size_t/int/long inconsistencies
- fix weak alias stuff
- make paths consistent
- fix config files
- use switch instead of if/then/elseif...
- fix Makefiles
exceptions now, there's an unplesant side effect that when FPSCR is
unavailable, your last display will get disabled just in case, "to
avoid infinite recursion". That happens directly in throw_exception,
so even catching that NOT_AVAILABLE_ERROR doesn't help.
Tweak the code a bit so that sh_analyze_prologue only reads FPSCR as
needed, when an FMOV instruction is encountered in the prologue.
XXX: I'm not sure if this is the right thing to do, but it seems
minimally intrusive.
discard old code brought over from gdb6.
Add regmap for FPU regs. We don't support it now but it's irrelevant
here (though we effectively commit to using the same layout as Linux
and OpenBSD).
Define GDB_OSABI_NETBSD_CORE like sparc does to be forward compatible
with upstream.
reality. Recognize old PT___GETREGS40 (for struct __reg40 without
GBR). Recognize PT_GETFPREGS that we don't yet actually have, but for
which the number is now reserved.
- The Postfix SMTP daemon sent "bare" newline characters instead of
<CR><LF> when a header_checks REJECT pattern matched multi-line
header. This bug was introduced with Postfix 1.1.
- The Postfix SMTP daemon sent "bare" newline characters instead of
<CR><LF> when an smtpd_proxy_filter returned a multi-line
response. This bug was introduced with Postfix 2.1.
- For compatibility with future EAI (email address
internationalization) implementations, the Postfix MIME processor no
longer enforces the strict_mime_encoding_domain check on unknown
message subtypes such as message/global*. This check is disabled by
default.
- The Postfix master daemon could report a panic error ("master_spawn:
at process limit") after the process limit for some service was
reduced with "postfix reload". This bug existed in all Postfix
versions.
it gets signal 9, which prints "Killed" -- that this is actually
signal 9 and not a generic message is never clear and causes people to
file PRs unnecessarily.
Suggested by followups to PR 44096.
- Support to output HTML fragments
- Initial support for mdoc->man translation
- Initial eqn support
- preconv / apropos / demandoc (not hooked up)
- Various bugfixes
trying to unwind while unwinding; plus the sniffer is not allowed to unwind.
- Return the outer frame to terminate not the null frame.
This fixes the kernel track unwinder, so enable it.
When -Bsymbolic, all resolved symbols need to be created with relative
relocations. The dynamic linker depends on that because it only
knows how to perform relative relocations in the first pass to
relocate itself. This bug ended up creating R_SPARC_64 symbols for
12 relocation symbols in ld.elf_so, one of them dlopen(), which
ended up breaking things using dlopen(), most notably PAM.
for platforms with shlibs already (and are.)
this obsoletes our hacks for the libgcc specs to use libgcc_pic, and
fixes a couple of other issues reported to me directly.
targets. Use this function in the initiator, to ensure that the same limit
is used throughout (i.e. stop you attempting to compile the initiator with a
different limit to the library - the initiator uses the library for most of
its functionality).
If truncating the number of targets in the initiator, truncate it to the
correct amount, not half of what it should be.
Obey RFC3720 when determining targets. SendTargets=All MUST be supported
on a discovery session, and MUST NOT be supported on an operational session.
Previously, SendTargets=All was used both in the initial discovery session
and the later session. (from Daisuke Aoyama - author of istgt).
xsrc/external/mit/xkeyboard-config/dist/ directory, i.e.
- pull changes from xkeyboard-config 2.0 to 2.3
- put back entries for sharp machines (zaurus and hpcarm wzero3s)
removed in rev 1.4 (xkeyboard-config 2.0 merge)
XXX1: can't we generate these files during build?
XXX2: hpcarm W-ZERO3 keyboard on X still has some troubles...
gdb asserts when a frame returns 0. In previous versions of gdb, it just
printed 0, and stopped scanning. Our unwinder produces the same results in
both versions of gdb, i.e. it is broken and worse than the current gdb
default. We will need to revisit this for kernel debugging.
__attribute__((__warn_unused_result__)) and the __builtin_expect()
use here is an attempt to silence the warning. This does not work
if the second argument of __builtin_expect() is not a constant,
but we are not linux, so instead of going through the
__builtin_constant_p() dance to correct this, just disable it for
non linux OS's.
