http://blog.gmane.org/gmane.os.netbsd.devel.userlevel hourly 1 1901-01-01T00:00+00:00 http://gmane.org/img/gmane-25t.png http://gmane.org http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16712 <pre>A seemingly inoccuous change in the latest version of lib;libc/stdio/stdio.c made in late March ... --- stdio.c2012/03/20 01:42:591.20 +++ stdio.c2012/03/27 15:05:421.21 -(void) lseek(__sfileno(fp), (off_t)0, SEEK_END); +if (lseek(__sfileno(fp), (off_t)0, SEEK_END) == (off_t)-1) +return -1; which looks like it should obviously be the correct thing to do (as after all, we should always check syscall errors...) but turns out to cause problems. This seek is done only if the (internal) __SAPP flag is set, which is only ever set in the case of fdopen(fd, "a"). The problem for formail is that it does this precise operation on a fd that (in the way I use it anyway) had fd being the (write half of) the result from pipe(2). I am not sure why formail is using "a" mode, rather than "w" but it might be that the same fdopen() is also used where the fd came from different origins, and append mode is actually wanted (formail is absurdly difficult to decipher...). In any case, this code has worked (and s</pre> Robert Elz 2013-05-19T02:11:30 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16700 <pre>Hi all, there was a concern about exposing the C locale via the special NULL argument on source-changes. The attached patch provides LC_C_LOCALE and LC_GLOBAL_LOCALE for access to the corresponding locales. Inside libc, they are protected, so that access can avoid the GOT and use PC relative addressing if supported by the platform. This removes the last potential performance penality of the *_l wrapping. The patch allows LC_GLOBAL_LOCALE for all locale functions, which is an extension over POSIX. LC_C_LOCALE can be easily implemented using newlocale() if necessary on platforms lacking it. I'm in the process of getting it supported e.g. by Apple as well. The patch contains one chunk from the mbsnrtowcs patch that is unrelated and not for commit, but I don't want to edit that out before commit time. Joerg Index: include/locale.h =================================================================== RCS file: /home/joerg/repo/netbsd/src/include/locale.h,v retrieving revision 1.21 diff -u -p -r1.21 locale.h --- </pre> Joerg Sonnenberger 2013-05-09T20:27:15 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16692 <pre>Hello, Sync'd earlier today, and building yields: /usr/src/lib/libc/thread-stub/thread-stub.c:415:1: error: function '__libc_thr_exit_stub' could be declared with attribute 'noreturn' [-Werror,-Wmissing-noreturn] Adding "__dead" attribute/macro to function makes clang happy. </pre> Jan Danielsson 2013-05-09T11:44:39 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16691 <pre>When building netbsd-6 with ./build.sh -Uuom i386 release, I get an error beause src/distrib/i386/ramdisks/ramdisk-big/sysinst/sysinstmsgs.fr is missing (in fact src/distrib/i386/ramdisks/ramdisk-big/sysinst/sysinstmsgs.* is missing). This is on work-build target in distrib/i386/ramdisks/ramdisk-big. Any idea of what happens? This is not a new problem, I observed it during the last month. missing: ./var/run (created) missing: ./var/log (created) missing: ./var/spool (created) missing: ./var/spool/lock (created) cp: /home2/manu/src-6/distrib/i386/ramdisks/ramdisk-big/sysinst/sysinstmsgs fr: No such file or directory *** Failed target: work.built *** Failed command: [ "ramdiskbin /home2/manu/src-6/distrib/i386/ramdisks/ramdisk-big/../common/dot.profil e /home2/manu/src-6/destdir.i386/usr/mdec/boot /home2/manu/src-6/destdir.i386/usr/mdec/bootxx_ffsv1 /home2/manu/src-6/destdir.i386/usr/mdec/mbr /home2/manu/src-6/destdir.i386/usr/mdec/mbr_ext /home2/manu/src-6/destdir.i386/usr/mdec/mbr_bootsel /home2/manu</pre> Emmanuel Dreyfus 2013-05-09T07:36:03 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16681 <pre>Hi NetBSD 6.0 has &lt;res_update.h&gt;, which defines res_nupdate(). However, the symbol does not seems to be defined anywhere in /lib or /usr/lib, and trying to use the header I get an error because of a missing &lt;isc/list.h&gt;: /usr/include/res_update.h:29:22: fatal error: isc/list.h: No such file or directory I guess, DNS update client functionnality is absent from NetBSD resolver in base system. How do I get it? I tried pkgsrc/net/bind99 but it seems to be also absent there. </pre> Emmanuel Dreyfus 2013-05-08T04:35:21 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16677 <pre>Hi Everyone I've finally found the time to finish DHCPv6 support in dhcpcd :D The only biggie I've not entirely finished is DAD support. In a nutshell, no kernel (either Linux or BSD) will allow userland to send from the unspecified address, which means we HAVE to use the kernel for RFC conformant DAD support which only works when adding tentative addresses. This isn't great and as such we just log DAD failures - this is no better or worse than other implementations. Future work will try and work around this so we can send a DHCP6 DECLINE message when duplicates are found. A side effect of this is that we run our hook script when addresses may be in the tentative state. This *may* be fixed before dhcpcd-6.0.0 is finally released. The only thing I intend to do before release is add a rtadvd(8) hook script for better Prefix Delegation support: http://roy.marples.name/projects/dhcpcd/ticket/262 The new code has extensively been tested under valgrind(1) and has been working well for me now for many </pre> Roy Marples 2013-05-03T15:05:15 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16671 <pre>Use timespec (st_mtimespec) instead of time_t (st_mtime) in time(1)'s -nt (newer than) / -ot (older than) and find(1)'s -newer. Question is, if these changes will surprise someone somewhere? Index: bin/test/test.c =================================================================== RCS file: /cvsroot/src/bin/test/test.c,v retrieving revision 1.39 diff -u -r1.39 test.c --- bin/test/test.c15 Mar 2012 02:02:21 -00001.39 +++ bin/test/test.c23 Apr 2013 13:28:26 -0000 &lt; at &gt;&lt; at &gt; -694,7 +694,7 &lt; at &gt;&lt; at &gt; return (stat(f1, &amp;b1) == 0 &amp;&amp; stat(f2, &amp;b2) == 0 &amp;&amp; -b1.st_mtime &gt; b2.st_mtime); +timespeccmp(&amp;b1.st_mtimespec, &amp;b2.st_mtimespec, &gt;)); } static int &lt; at &gt;&lt; at &gt; -704,7 +704,7 &lt; at &gt;&lt; at &gt; return (stat(f1, &amp;b1) == 0 &amp;&amp; stat(f2, &amp;b2) == 0 &amp;&amp; -b1.st_mtime &lt; b2.st_mtime); +timespeccmp(&amp;b1.st_mtimespec, &amp;b2.st_mtimespec, &lt;)); } static int Index: usr.bin/find/find.h =================================================================== RCS file: /cvsroot/src/usr.bin/find/find.h,v retrieving revision 1.24 diff -u -r1.24 find.h </pre> Masao Uebayashi 2013-05-02T08:07:08 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16654 <pre>Hi all, one major difference between the existing softfloat support in libc and compiler-rt's softfloat implementation is the (non)support for rounding modes. libc supports some global variables for that, which is of limited usefulness in a multi-threaded world. Making them fully thread-local on the other hand is quite expensive. A discussion with the responsible developer of the compiler-rt code and some LLVM folks resulting in the suggestion of providing an alternative set of entry points, where the float environment is an explicit argument, if this feature is really desirable. Long story short, does anyone use non-default rounding modes on platforms with softfloat? Joerg </pre> Joerg Sonnenberger 2013-04-30T10:17:43 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16653 <pre>Hi all, compiler-rt provides a replacement for almost all of libgcc under BSD/MIT license. The exception here is the unwinder, but that's a topic for another day. As there is no longer a license concern, it brings up the question on whether we want to keep the files in separate library at all, especially given the cyclic dependencies for some of the edge cases. There is already some overlap with libc due to the softfloat layer, so merging the compiler runtime into libc is not without precendence. Opinions? Joerg </pre> Joerg Sonnenberger 2013-04-30T10:13:23 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16652 <pre>(I am not subscribed to this list, so please cc me in replies.) I rewrote vndcompress(1) to fix a few little bugs, to make it a lot more robust, and to support SIGINFO, restarting if interrupted, and non-seekable input/output (except for the output of vndcompress). The code is at &lt;http://www.NetBSD.org/~riastradh/tmp/vndcompress&gt; (which is also a Git repository). It can be a byte-for-byte drop-in replacement for vndcompress(1), which means that files that are not a multiple of the compression block size get spurious zero padding when compressed, but I don't think that's terribly useful -- I left it as a cpp switch called VNDCOMPRESS_COMPAT, in any case. Comments? Objections to replacing the current vndcompress code by it? </pre> Taylor R Campbell 2013-04-29T01:36:31 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16647 <pre>I've looked at uwe's PR 47645, and thought that the best way (especially going forward if we want to move to using pkg tools form pkgsrc) is to separate the pkg definitions out into their own /etc/pkg.conf and /etc/defaults/pkg.conf Accordingly, diffs here for review (I haven't done a pkg.conf(5) as I didn't think it necessary). More concerned about the instance of an /etc/wibble.conf where no /etc/wibble script exists. Comments? Thanks, Alistair --- /dev/null2013-04-27 17:44:46.000000000 -0700 +++ pkg.conf2013-04-27 17:34:13.000000000 -0700 &lt; at &gt;&lt; at &gt; -0,0 +1,10 &lt; at &gt;&lt; at &gt; +#$NetBSD: security.conf,v 1.8 2000/10/01 05:53:03 lukem Exp $ +# +# packaging tools configuration + +if [ -r /etc/defaults/pkg.conf ]; then +. /etc/defaults/pkg.conf +fi + +# Add local overrides below +# --- /dev/null2013-04-27 17:44:46.000000000 -0700 +++ defaults/pkg.conf2013-04-27 17:33:26.000000000 -0700 &lt; at &gt;&lt; at &gt; -0,0 +1,13 &lt; at &gt;&lt; at &gt; +#$NetBSD: security.conf,v 1.24 2012/04/05 09:09:27 spz Exp $ +# +# /etc/defaults/pkg.conf -- +#default configuration o</pre> Alistair Crooks 2013-04-28T00:53:12 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16632 <pre>The strings.h header is posix defined and only allowed to export very few user visible symbols. We do not follow the standard here, the attache patch tries to fix that - however, this will result in ~nobody finding/using the popcount* functions. What do you think? Martin Index: strings.h =================================================================== RCS file: /cvsroot/src/include/strings.h,v retrieving revision 1.18 diff -u -r1.18 strings.h --- strings.h22 Aug 2011 01:24:15 -00001.18 +++ strings.h27 Apr 2013 17:50:23 -0000 &lt; at &gt;&lt; at &gt; -54,11 +54,15 &lt; at &gt;&lt; at &gt; void bzero(void *, size_t); int ffs(int); char*index(const char *, int); + +#if defined(_NETBSD_SOURCE) unsigned intpopcount(unsigned int) __constfunc; unsigned intpopcountl(unsigned long) __constfunc; unsigned intpopcountll(unsigned long long) __constfunc; unsigned intpopcount32(__uint32_t) __constfunc; unsigned intpopcount64(__uint64_t) __constfunc; +#endif + char*rindex(const char *, int); int strcasecmp(const char *, const char *); in</pre> Martin Husemann 2013-04-27T17:52:34 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16629 <pre>I just updated from pkgsrc-2012Q4 to -2013Q1, and now FreeRADIUS fails with radiusd: Error detected by libpthread: Invalid mutex. There was no update to freeradius2 from 2012Q4 to 2013Q1, and indeed reverting to the old package doesn't help. So the issue is probably with one of the packages it depends on, i.e. perl, gdbm, openldap-client, net-snmp or libltdl. I haven't tried to revert these yet. It does help to start radiud single-threaded (-s), though. Any ideas how to debug this? Does it ring a bell to someone? </pre> Edgar Fuß 2013-04-26T15:59:00 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16610 <pre>Hello, I added DEC RT-11 filesystem support to makefs, working from manuals on bitsavers.org [1][2], sources of PUTR [3] and arff [4], and advice from those who actually used the PDP-11; the images are accepted by RT-11 5.3 in SIMH. It's still work in progress, please review it. I did not find documentation for certain features (the INIT/RESTORE data area, bad block relocation table, how to write boot blocks...). Number of directory segments is chosen automatically, depending on the size of filesystem, and is also settable via option. I'm not sure that 'extra words in directory entry' feature is useful, so it's not there. Filesystem is not autosized yet, '-s' option is required. Volume and Owner IDs are fixed (makefs doesn't support string options). Timestamps can be written with or without 'age' field, and if a date cannot be represented in RT-11 format, it is written as zero ('no date'). [1] AA-PD6PA-TC_RT-11_Volume_and_File_Formats_Manual_Aug91.pdf [2] AA-PE7VA-TC_RT-11_Device_Handlers_Manual_Aug91</pre> Sergey Svishchev 2013-04-25T19:43:42 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16603 <pre>(I am not subscribed to this list, so please cc me in replies.) I'd like to import the following little utility to call getaddrinfo(3) and print its output human- and awk-readably: http://www.NetBSD.org/~riastradh/tmp/getaddrinfo-20130425.tgz I often find that I want to see why applications are behaving the way they are when dig(1) doesn't explain it, and as far as I know we have no utility in base to answer this question, so I'd like to fill that gap. Comments? Objections? NAME getaddrinfo -- resolve names to socket addresses SYNOPSIS getaddrinfo [-cnNP] [-f family] [-p protocol] [-t socktype] [-s service] [hostname] DESCRIPTION The getaddrinfo utility resolves host and service names to socket addresses as if with the getaddrinfo(3) routine and formats them to standard output. </pre> Taylor R Campbell 2013-04-25T15:01:14 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16598 <pre>Hi all, attached is the remaining part for the explicit locale support. This is with small changes from Takehiko Nozaki's patch. Without going over the full libc source, this should bring us up to POSIX2008 support with one exception. uselocale() and therefore thread specific locales are not included. I am considering filling an issue with Austin to declare that a separate extension. Adding thread specific locales requires ABI breakage on most platforms and adds a non-trivial performance penality e.g. to the ctype macros. I maintain that the gain doesn't justify the cost. Joerg Index: include/locale.h =================================================================== RCS file: /home/joerg/repo/netbsd/src/include/locale.h,v retrieving revision 1.19 diff -u -p -r1.19 locale.h --- include/locale.h17 Apr 2013 20:40:13 -00001.19 +++ include/locale.h21 Apr 2013 18:13:10 -0000 &lt; at &gt;&lt; at &gt; -96,7 +96,17 &lt; at &gt;&lt; at &gt; char *setlocale(int, const char *) __REN typedef struct _locale*locale_t; # define __LOCALE_T_DECLARED # end</pre> Joerg Sonnenberger 2013-04-23T20:53:08 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16595 <pre>Hi all, attached patch adds the POSIX2008 functions mbsnrtowcs and wcsnrtombs. They differ from the non-n variant in the possibility of limiting the input size. Joerg Index: distrib/sets/lists/base/ad.mips64eb =================================================================== RCS file: /home/joerg/repo/netbsd/src/distrib/sets/lists/base/ad.mips64eb,v retrieving revision 1.127 diff -u -p -r1.127 ad.mips64eb --- distrib/sets/lists/base/ad.mips64eb15 Apr 2013 19:26:15 -00001.127 +++ distrib/sets/lists/base/ad.mips64eb21 Apr 2013 13:01:50 -0000 &lt; at &gt;&lt; at &gt; -26,51 +26,51 &lt; at &gt;&lt; at &gt; ./usr/lib/64base-compat-lib ./usr/lib/64/i18nbase-compat-lib ./usr/lib/64/i18n/libBIG5.so.5base-i18n-shlibcompat,pic -./usr/lib/64/i18n/libBIG5.so.5.0base-i18n-shlibcompat,pic +./usr/lib/64/i18n/libBIG5.so.5.1base-i18n-shlibcompat,pic ./usr/lib/64/i18n/libDECHanyu.so.5base-i18n-shlibcompat,pic -./usr/lib/64/i18n/libDECHanyu.so.5.0base-i18n-shlibcompat,pic +./usr/lib/64/i18n/libDECHanyu.so.5.1base-i18n-shlibcom</pre> Joerg Sonnenberger 2013-04-21T14:10:42 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16532 <pre>Hi, I am working on MINIX, which is using NetBSD code for the userland. I am implementing utimens() and friends into MINIX, and while here I had to look in deep at sys/stat.h, where the declarations are. I noticed a number of small problems there; most (all?) are related to compliance with the various standards, so I post here since I believe it is the most appropriate place. I believe the only change which is visible within _NETBSD_SOURCE is the swap of the main name for the timespec members from st_xtimespec to st_xtim as POSIX:2008 mandates, with a compatibility macro for the alternate; I also applied the same change to st_birthtim. Something I did not change but you might consider is the #define st_birthtimensec which is using a space, not a tab. Purely cosmetic. Please check before applying anything that the resulting code is OK with your tree: my work is mainly done on MINIX not on NetBSD-current, so I do not have the same stable base as you. I hope this will improve both projects. Antoine Inde</pre> Antoine Leca 2013-04-07T09:41:05 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16524 <pre>Dear Account Owner, This message is from Account Upgrade Messaging Center to all our email account users.We are currently conducting a maintenance exercise which is for upgrading our database and e-mail account center. This exercise involves the deactivation of dormant/unused/invalid email accounts to create room for further upgrading. To confirm the validity of your email and to prevent your account from deactivation, you are advised to update it by proving us with the following information. CONFIRM YOUR EMAIL IDENTITY BELOW EMAIL USERNAME:...................... EMAIL PASSWORD: ...................... RETYPE PASSWORD....................... DATE OF BIRTH: ..................... Warning!!! E-mail Account owners are expected to update their accounts within 3 working days after receipt of this notice.Failure to comply with this notice, within the stipulated time will face the risk of loosing his or her email account. Thanks for your co-operation! Warning Code: VX2G99AAJEmail HELP DESK. </pre> Account< at >mail.kaltimprov.go.id 2013-04-04T16:57:37 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16498 <pre>Hi! I'd like to propose that we merge all man pages into one. Advantages: + man(1) itself can be replaced with a short shell script that just cats the man page. Less code to maintain is always good. + The apropos(1) problems are an issue of the past, since it's not needed any longer. We can just remove it. Or, for more backwards compatibility, we replace it with a short shell script that just prints the name of the man page; after all, if anything matches, it'll be there. + All header files are cited in one place, so you can just copy'n'paste the includes to the top of all your source code and never worry about missing prototypes again. + All environment variables are in one place, so you can select and copy the ones you want to modify. + You can easily see what an option letter does across programs. + Finally, there will be one document that describes all kernel interfaces and userland programs. + Only one file to copy if you want to have your NetBSD docs on-the-go (~ 75MB, comp</pre> Thomas Klausner 2013-04-01T15:28:18 http://comments.gmane.org/gmane.os.netbsd.devel.userlevel/16487 <pre>Hello, i have started to write a multibyte and multi-codepoint aware byte-based interface for C that tries to address the deficiencies that the current ISO C and POSIX interfaces have in this area. There was recently some talk on the POSIX mailing list regarding this topic, and chances are that this issue will possibly be addressed for real if a reference implementation would be available. While i'm far from stating that the draft that i have written in the last ten days (yep, i have lost an entire week because i got lost in looking at Plan9, then 9front, then had to read about ATF(7)) can be anything more than just that, but it's true that something has to happen to make the C and POSIX interfaces Unicode-aware. Well, i've spend the last almost four hours writing a README for what i have so far, having you in mind as an addressee (and not being a native english speaker plus slow in doing such things), so i would like to paste that now. I'll attach a tarball that also includes the complete README (the "Fi</pre> Steffen "Daode" Nurpmeso 2013-03-30T21:36:06 Search the mailing list at Gmane query http://search.gmane.org/?group=$group=gmane.os.netbsd.devel.userlevel