Re: start building clang alongside gcc on amd64

On Mon, Apr 17, 2017 at 09:20:21AM +0200, Mark Kettenis wrote: > After further discussion it Theo, here is a new version of the diff. > It explicitly lists the gcc4 architectures now, introduces BUILD_GCC3 > and BUILD_GCC4 variables and fixes installation of the c++ headers. > > Index:

Re: start building clang alongside gcc on amd64

> Date: Mon, 17 Apr 2017 09:26:53 +0200 > From: Landry Breuil > Content-Disposition: inline > List-Owner: > X-Loop: tech@openbsd.org > Sender: owner-t...@openbsd.org > X-CNFS-Analysis: v=2.2 cv=eoad9chX c=1 sm=0 tr=0 >

Re: armv7/omap: introducing amdisplay(4) + nxptda(4)

xOn Sun, Apr 16, 2017 at 12:07:34AM -0400, Ian Sutton wrote: > > The USB host controller on the am335x isn't EHCI compatible and is currently > unsupported, the only direct interface with the board is via a com(4) ftdi > serial port that doesn't provide a wskbd(4) device which Xorg demands. None

Re: Fix regress/lib/libc/db

Maybe use UINT_MAX since st_size gets cast to u_int below? Either way OK millert@ - todd

Re: Fix regress/lib/libc/db

> From: "Todd C. Miller" > Date: Mon, 17 Apr 2017 08:49:26 -0600 > > Maybe use UINT_MAX since st_size gets cast to u_int below? > Either way OK millert@ Already committed. But INT_MAX is better anyway since there is a cast to int as well and INT_MAX < UINT_MAX.

softraid: Fixup concat

Y'all, During some refactoring work, the volume size calculation for the concat discipline got lost, resulting in zero sized volumes. http://secnorth.net/diffs/softraid-concat.diff: diff --git a/sys/dev/softraid_concat.c b/sys/dev/softraid_concat.c index 90cb12c8b6e..11fbf8ddb3f 100644 ---

Re: tcpdump: decode BGP Administrative Shutdown Communication

Hi OpenBSD, bgpd(8) as shipped in OpenBSD 6.1 supports draft-ietf-idr-shutdown-07. The below patch adds support to tcpdump(8) to decode such shutdown communication. This is an improved version of the patch proposal I sent in January. Kind regards, Job diff --git

Use correct libressl version in pkg-config

The pkg-config files for libcrypto and libssl claim to be version 1.0.0. This was noticed on the openvpn-devel mailing list when they started requiring OpenSSL 1.0.1 or newer: https://sourceforge.net/p/openvpn/mailman/message/35785095/ While I don't think their approach of testing specific

Re: Fix regress/lib/libc/db

On Mon, 17 Apr 2017 17:12:13 +0200, Mark Kettenis wrote: > > From: "Todd C. Miller" > > Date: Mon, 17 Apr 2017 08:49:26 -0600 > > > > Maybe use UINT_MAX since st_size gets cast to u_int below? > > Either way OK millert@ > > Already committed. But INT_MAX is better

Re: start building clang alongside gcc on amd64

After further discussion it Theo, here is a new version of the diff. It explicitly lists the gcc4 architectures now, introduces BUILD_GCC3 and BUILD_GCC4 variables and fixes installation of the c++ headers. After applying this diff you need to: 1. Run make install in /usr/src/share/mk 2.

sparc64 libunwind

The diff below adds sparc64 support to libunwind. With this diff I can build libc++abi and libc++ on sparc64. The exception handling regress tests in src/regress/misc pass when compiled with clang/clang++. The sparc64 unwinder is a little bit more complicated than the others since it has to

Re: start building clang alongside gcc on amd64

On Mon, Apr 17, 2017 at 10:04:47AM +0200, Mark Kettenis wrote: > > Date: Mon, 17 Apr 2017 09:26:53 +0200 > > From: Landry Breuil > > Content-Disposition: inline > > List-Owner: > > X-Loop: tech@openbsd.org > > Sender: owner-t...@openbsd.org > >

Re: start building clang alongside gcc on amd64

On Mon, 17 Apr 2017 09:20:21 +0200 (CEST) Mark Kettenis wrote: > These instructions may not be 100% accurate yet ;). I also had to build libcxx before doing a full build: # cd /usr/src/lib/libcxx # make obj # make depend # make includes # make # make install

Re: ld.lld: disable colors/bold/whatever

> Date: Sun, 16 Apr 2017 22:00:23 +0200 (CEST) > From: Mark Kettenis > > > Date: Sun, 16 Apr 2017 21:21:24 +0200 > > From: Patrick Wildt > > > > Hi, > > > > this diff disables the color diagnostics by default so that LLVM's lld > > does not throw

Re: ld.lld: disable colors/bold/whatever

On Mon, Apr 17, 2017 at 06:51:57PM +0200, Pascal Stumpf wrote: > On Mon, 17 Apr 2017 18:25:34 +0200 (CEST), Mark Kettenis wrote: > > > Date: Sun, 16 Apr 2017 22:00:23 +0200 (CEST) > > > From: Mark Kettenis > > > > > > > Date: Sun, 16 Apr 2017 21:21:24 +0200 > > > > From:

sshd: Remove authorized_keys2 file

Now that protocol version 1 was finally dropped in sshd(8), get rid of this file completely. Our default sshd_config(5) overwrites AuthorizedKeysFile to ignore it anyway and sshd(8)'s FILES section doesn't mention it either. Index: etc/changelist

Re: ld.lld: disable colors/bold/whatever

> From: Pascal Stumpf > Date: Mon, 17 Apr 2017 18:51:57 +0200 > > On Mon, 17 Apr 2017 18:25:34 +0200 (CEST), Mark Kettenis wrote: > > > Date: Sun, 16 Apr 2017 22:00:23 +0200 (CEST) > > > From: Mark Kettenis > > > > > > > Date: Sun, 16 Apr 2017

Re: ld.lld: disable colors/bold/whatever

On Mon, 17 Apr 2017 18:25:34 +0200 (CEST), Mark Kettenis wrote: > > Date: Sun, 16 Apr 2017 22:00:23 +0200 (CEST) > > From: Mark Kettenis > > > > > Date: Sun, 16 Apr 2017 21:21:24 +0200 > > > From: Patrick Wildt > > > > > > Hi, > > > > > > this diff

Re: ld.lld: disable colors/bold/whatever

> From: Pascal Stumpf > Date: Mon, 17 Apr 2017 18:51:57 +0200 > > On Mon, 17 Apr 2017 18:25:34 +0200 (CEST), Mark Kettenis wrote: > > > Date: Sun, 16 Apr 2017 22:00:23 +0200 (CEST) > > > From: Mark Kettenis > > > > > > > Date: Sun, 16 Apr 2017

Re: tcpdump: decode BGP Administrative Shutdown Communication

Hi all, Daan Keuper (Computest) was kind enough to review the diff, he pointed out the following: safeputs() expects a null-terminated string. Since shutdown_comm_length won't exceed BGP_NOTIFY_MINOR_CEASE_ADMIN_SHUTDOWN_LEN (128), the following will ensure a null-terminated string is passed to

Re: tcpdump: decode BGP Administrative Shutdown Communication

+ memset(string, 0, 129); + memcpy(string, p+1, shutdown_comm_length); + safeputs(string); Please don't copy numbers like that. If this is a string, why not use string functions that gaurantee truncation

vscsi_t2i() unused variable

Hello, The variable rv in vscsi_t2i() appears to always be zero and could be removed. - Michael Index: src/sys/dev/vscsi.c === RCS file: /cvs/src/sys/dev/vscsi.c,v retrieving revision 1.41 diff -u -p -u -r1.41 vscsi.c ---

delete pf states by exact flow info and speed it up

Hi, I'd like to add "key" modifier to "pfctl -k" to delete pf states, like below: # pfctl -k key -k 'tcp 10.0.0.1:80 <- 10.0.0.101:32123' This deletion will be very fast if the RB tree of pf state table is used to find the state. The diff also includes the diff for that part (DIOCKILLSTATES

Re: CFI directives support in binutils

On Mon, Apr 17, 2017 at 10:45:37PM +0200, Mark Kettenis wrote: > This diff extends the support for CFI in binutils with a few > directives emitted by clang. Most of the code is stolen from > FreeBSD's GPLv2 binutils. Support for the .cfi_sections directive is > incomplete; basically we just

CFI directives support in binutils

This diff extends the support for CFI in binutils with a few directives emitted by clang. Most of the code is stolen from FreeBSD's GPLv2 binutils. Support for the .cfi_sections directive is incomplete; basically we just ignore it. Our binutils only support emitting frame info in the .eh_frame