-changes-d/2014/11/23/msg007400.html
Taylor claims the existing abcksum() also violates aliasing rule.
http://mail-index.netbsd.org/source-changes-d/2014/11/23/msg007402.html
If he is correct it's no sense to tweak only functions complained
by current gcc48.
---
Izumi Tsutsui
or memcpy().
Then you don't have any patch for existing (and not warned) abcksum().
Are you also okay to specify -fno-strict-aliasing with the original code
as I and mrg (and now Taylor) suggest, rather than patching only
warned assignments?
---
Izumi Tsutsui
christos@ wrote:
On Nov 24, 7:18am, tsut...@ceres.dti.ne.jp (Izumi Tsutsui) wrote:
| Then you don't have any patch for existing (and not warned) abcksum().
| Are you also okay to specify -fno-strict-aliasing with the original code
| as I and mrg (and now Taylor) suggest, rather than patching
as they come
along. If we did not do that we would still be in the age of KR. Just
because a situation is not perfect we don't aim to make it worse.
Your strategy is still best or nothing.
My goal is acceptable compromise because
we have fewer resources than old days.
That's all.
---
Izumi Tsutsui
TierII users must follow core's
(or other non-user developer's) non-essencial claims
about dumb MD source implementation, no one will maintain
such ports.
---
Izumi Tsutsui
.
I've been really tired and lost my motivation.
Sorry.
---
Izumi Tsutsui
image. Then you
can test.
Does this sound reasonable?
Sounds unlikely for netbsd-7.
I'm afraid you don't read how the MD installboot.c does weird operations
at all.
---
Izumi Tsutsui
host16enc and target16enc for streaming data, but the target
cksum is not stream but just calculated in uint16_t.
---
Izumi Tsutsui
are shown in a patch to appease compiler
makes no sense even for correctness.
---
Izumi Tsutsui
()
#1 0x0006f9b2 in compile_stream.clone ()
#2 0x0006fbf2 in compile$$from$$sed ()
#3 0x0006ff28 in _crunched_sed_stub ()
#4 0x21e6 in ___start ()
#5 0x20d4 in _start ()
(gdb)
---
Izumi Tsutsui
0x00063010 in compile () at /r/work/netbsd-7/src/usr.bin/sed/compile.c:144
#5 0x00101a24 in main (argc=optimized out, argv=0xffefecc0)
at /r/work/netbsd-7/src/usr.bin/sed/main.c:207
#6 0x21e6 in ___start ()
#7 0x20d4 in _start ()
(gdb)
---
Izumi Tsutsui
? 0
: 1;
Ok, this is a bug in libhack; please cvs update.
Christos, could you please at least compile before commit?
http://mail-index.netbsd.org/source-changes/2014/11/15/msg060599.html
Log Message:
Remove unused variable.
Such botch and additional fixes make pullups annoying...
---
Izumi
, unexpected bugs could often slip in by such untested
mechanical changes and it makes harder to track problems in future.
I'm still trying to get atari booting with HAVE_GCC=48,
but there are still more problems...
(sysinst floppy overflow, sed(1) in the install floppy dumps core, etc)
---
Izumi Tsutsui
trigger it?
---
Izumi Tsutsui
christos@ wrote:
On Nov 14, 3:00am, tsut...@ceres.dti.ne.jp (Izumi Tsutsui) wrote:
| Test it (or call for testers) before commit
| because installboot could be fatal on install floppy and bootstrap.
|
| but memcpy is the only way.
|
| - cast via (void *)
That does not work
and the a.out hack seems a special case.
---
Izumi Tsutsui
of obvious in the GCC Project is: could not possibly
cause anyone to object. We adopt this definition here)
All other (i. e. non-obvious) fixes should have a review.
It looks there are so many blames for your recent changes.
---
Izumi Tsutsui
(u_int16_t *)bb-bb_xxboot + 255
(the cksum word looks at the end of 512 byte sector).
memcpy(9) looks also awful for readers because it hides endianness..
---
Izumi Tsutsui
bootloader though).
I wonder if you need a different script (or proper wrapper)
for modules..
---
Izumi Tsutsui
.
Anyway, it looks required for a.out only but no one will fix
a.out features..
---
Izumi Tsutsui
a look at m68k gcc stuff?
http://releng.netbsd.org/builds/HEAD/201408251020Z/hp300.build.failed
m68k--netbsdelf-gcc:
/home/source/ab/HEAD/src/external/gpl3/gcc.old/lib/libgcc/libgcc_eh/../arch/m68k.mk:
linker input file unused because linking not done
Thanks,
---
Izumi Tsutsui
/builds.cgi
Please fix or revert.
---
Izumi Tsutsui
which have size restrictoin in bootloaders,
for example.
If you will work on it yourself no one will complain,
but your comments look like as an armchair philosopher.
---
Izumi Tsutsui
sense to dd 100K of null data to a file then delete in
the build to ensure there is enough space in the final image?
Probably we could specify
IMGMAKEFSOPTIONS= -o bsize=4096,fsize=512 -b [proper free-blocks]
in each ramdisk Makefiles to stop build before redzone?
---
Izumi Tsutsui
target binary and crunched binary is annoying)
- x86 ports already have USB install images without ramdisk and
they are also useful enough for debug
---
Izumi Tsutsui
-default settings on choosing installation sets etc.
on ports which don't use tmpfs or mfs during installation.
(100KB would be enough for this newsmips case though)
---
Izumi Tsutsui
have
proper native maintainers should be removed before
the next release, as paranoiac people did on www.
There is no mechanism/rule to see
which messages are translated or not.
---
Izumi Tsutsui
.8dc3e4a80ea6bfb94b5a5334dbda9e5c56fcddea
Are they no longer required for new playstation2 port?
---
Izumi tsutsui
joerg@ wrote:
Log Message:
Apply OpenBSD src/sys/dev/pci/if_wpireg.h 1.17.
Can we get a better commit message in the future?
Our commit guideline also says Give proper credit.
http://www.netbsd.org/developers/commit-guidelines.html
---
Izumi Tsutsui
joerg@ wrote:
On Tue, Jul 01, 2014 at 10:16:38PM +0900, Izumi Tsutsui wrote:
joerg@ wrote:
Log Message:
Apply OpenBSD src/sys/dev/pci/if_wpireg.h 1.17.
Can we get a better commit message in the future?
Our commit guideline also says Give proper credit.
http
the kernel on it to speed up the boot process.
XXX The /boot bug should nevertheless be fixed.
FYI, gunziping a kernel on m68k machines (luna68k and x68k) is
also extremely slow (takes a few minutes) so probably it's
an MI libsa cread problem.
---
Izumi Tsutsui
and mail is gid 1002 in my host)
---
Izumi Tsutsui
on install CDs
I wonder if we should also prepare /var/tmp (in etc.rc scripts).
---
Izumi Tsutsui
from FreeBSD's soundcard.h, hopefully providing
some binary compatibility.
If new ioctls are added, shouldn't we bump at least minor in shlib_version
to denote API additions?
---
Izumi Tsutsui
skrll@ wrote:
How about the attached?
If it's tested on the target CPUs using mips/bus_dma.c,
I have no objection.
(the orignal code was tested only on R4400/R5000/Rm5200)
---
Izumi Tsutsui
Should be simple for you to copy across the optimisation then :)
Sorry, I have no motivation for current mips ports.
---
Izumi Tsutsui
skrll@ wrote:
I'm still seeing issues with my cobalt, but I'll keep hunting.
Probably all mips port MD bus_dma.c
(in arc, cobalt, emips, ews4800mips, hpcmips, mipsco, pmax, sgimips)
need the similar fix as mips/bus_dma.c rev 1.28.
(only algor, evbmips and sbmips use mips/bus_dma.c)
---
Izumi
they should also switch to the common mips bus space/dma files.
- does it handle MIPS1?
- BUS_DMASYNC_PREREAD seems incomplete
---
Izumi Tsutsui
skrll@ wrote:
yeah, they'll need that fix, but there are still cache aliasing bugs in
the pmap, I think.
PR/45746 is still open, btw.
---
Izumi Tsutsui
with gcc 4.5 and 4.8 (and same binaries are generated)
so I'll put it back to the Makefile.
---
Izumi Tsutsui
by 68000/010.
(and probably faster on 020 and later?)
---
Izumi Tsutsui
on mvme68k 060 kernels with both gcc 4.5 and 4.8.
In this boot_ustar case, passing only -Wa,-mcpu=m68000 fails with gcc 4.5,
and passing only -Wa,-march=m68000 (or -Wa,-m68030) fails with gcc 4.8.
---
Izumi Tsutsui
make a swap file?
Swap space in the distributed image is just annoying
for users to rawwrite it into devices.
To use resize_ffs(8), users have to edit disklabel anyway,
so I'm using a dumb script that checks disk size, updates labels,
and executes resize_ffs(8) for my custom images.
---
Izumi
update man page? Thanks.
---
Izumi Tsutsui
On Tue, Mar 04, 2014 at 12:11:59PM +, Izumi Tsutsui wrote:
Module Name:xsrc
Committed By: tsutsui
Date: Tue Mar 4 12:11:59 UTC 2014
Modified Files:
xsrc/external/mit/xorg-server/dist/hw/netbsd/x68k: x68kConfig.c
x68kGraph.c
Log
CRUNCHGEN_FLAGS
(several mips ports already have it), not in the common Makefile.
---
Izumi Tsutsui
settings. At least it should be set via
CRUNCHGEN_FLAGS?= -d ${DBG}
in the common Makefile.crunch.
Or properly remove all redundant CRUNCHGEN_FLAGS which specify -d
from existing Makefiles if you really don't like it.
---
Izumi Tsutsui
been annoying for me
for 15 years, but see no alternative.
---
Izumi Tsutsui
race.
---
Izumi Tsutsui
and it should not require installed DESTDIR.
For kernel builds, config(1) creates symlinks in compile dirs.
---
Izumi Tsutsui
the following messages?
http://mail-index.netbsd.org/tech-kern/2014/01/15/msg016421.html
http://mail-index.netbsd.org/tech-kern/2014/01/15/msg016425.html
---
Izumi Tsutsui
Module Name: src
Committed By: christos
Date: Wed Jan 8 02:15:42 UTC 2014
Modified Files:
src/lib/libc/stdlib: Makefile.inc ptsname.3 pty.c
Log Message:
add ptsname_r
Needs libc minor bump?
---
Izumi Tsutsui
~nothing uses it.
Then why do you add it if it will be never used?
---
Izumi Tsutsui
On Jan 8, 8:52pm, tsut...@ceres.dti.ne.jp (Izumi Tsutsui) wrote:
-- Subject: Re: CVS commit: src/lib/libc/stdlib
| Then why do you add it if it will be never used?
It will be never used != is not currently being used.
Then you must bump minor otherwise future binaries will fail silently
pgoyette@ wrote:
Module Name: src
Committed By: pgoyette
Date: Wed Jan 1 15:22:33 UTC 2014
Modified Files:
src/distrib/sets/lists/modules: mi
Log Message:
Add new crypto algorithm modules to sets list
You also have to update md.evbppc (for OEA/4xx/ebook).
---
Izumi
christos@ wrote:
On Jan 2, 1:25pm, tsut...@ceres.dti.ne.jp (Izumi Tsutsui) wrote:
-- Subject: Re: CVS commit: src/sys/arch/luna68k/dev
| This could lead to problems in the future. I would put it back.
|
| - it was added by me recently
| - netbsd-6 doesn't have it
| - no other
- CFLAGS needs -mc68000 (or -m68000) for sun2 binaries
- AFLAGS needs -Wa,mc68020 because SRT0.S includes 020 %cacr instructions
inside sun3 conditinal blocks
---
Izumi Tsutsui
access to the gcc-4.8.1 repo
so you can fix it yourself, just send me an ssh public key.
Why isn't it in the public branch?
---
Izumi Tsutsui
to have proper announcement in that case.
(even developers don't know what's ongoing)
---
Izumi Tsutsui
jnemeth@ wrote:
On Nov 8, 6:01am, Izumi Tsutsui wrote:
} christos@ wrote:
}
} | If you want, I can give you access to the gcc-4.8.1 repo
} | so you can fix it yourself, just send me an ssh public key.
} |
} | Why isn't it in the public branch?
}
} We did not want to burden
in the exec path.
Is there any exported documents that mention NetBSD specific ELF note?
---
Izumi Tsutsui
martin@ wrote:
On Tue, Nov 05, 2013 at 11:44:22PM +0900, Izumi Tsutsui wrote:
Is there any exported documents that mention NetBSD specific ELF note?
I am not aware of any (if you don't count comments in the exec_elf.h file).
Readelf/binutils seem to know about the .note.netbsd.ident one
about it on Twitter/Facebook etc.)
while it took too long..
---
Izumi Tsutsui
post proper announcements if you have fixed
dumb but fatal bugs per reports on public lists.
It's so annoying for users to track problems without descriptions.
Thanks,
---
Izumi Tsutsui
strncpy.S
Log Message:
Adjust for Coldfire
These changes break traditional m68k ports.
(struct uvm seems corrupted probably memset etc and kernel doesn't boot)
Please revert and create a branch for coldfire
if you can't debug your changes on m68k.
---
Izumi Tsutsui
(or ask x68k guys)?
At least dmesg(8) is mandatory for many developers.
Anyway, I guess the overflow was caused by recent disklabel(8) changes,
so I wonder if we can have SMALLPROG version of disklabel(8),
which supports only one $MACHINE (or $MACHINE_ARCH) like installboot(8).
---
Izumi Tsutsui
christos@ wrote:
From the amount of stuff that I had to fix in order to get to the point
to build in -current it seems that nobody has bothered to look at the
port for a while.
Someone[tm] should write a dumb script to post autobuild failure
messages (weekly?) to port-foo@ lists...
---
Izumi
/machine now points to usr/include/arm
How does MD disklabel.h work?
At least hpcarm and zaurus don't use arm/disklabel.h.
---
Izumi Tsutsui
\
+ -o d=4096 \
+ -b $((${extra}))m \
-F $tmp/selected_sets ${image} ${release} ${mnt}
fi
---
If the problem is overflow of the image, shouldn't the extra variable
in each config file be bumped instead of changing a unit?
---
Izumi Tsutsui
there's a better way to fix it, please do.
What actual problem did you get?
---
Izumi Tsutsui
pthread_condattr_setclock(3).
---
Izumi Tsutsui
the nature of the failure, would
not have fixed the problem.
Is it generally acceptable to bump minor
(i.e. adding new public functions in libs)
in release branches?
I'm afraid not all third party programs use proper configure script.
---
Izumi Tsutsui
ticket #722 (pthread_condattr_setclock(3) addition)?
Is that function really necessary for 6.1?
---
Izumi Tsutsui
)
---
Izumi Tsutsui
?
Last time I tried, it failed because NetBSD's rpc/types.h defined
enum_t and src/usr.bin/xlint/lint1/lint1.h for tools lint(1) also
defined different enum_t.
---
Izumi Tsutsui
?
Not really. For now Xorg uses PCVT-specific ioctl()s to switch virtual
consoles, until someone changes that to use wsdisplay we need PCVT for
that. ( and COMPAT_PCVT in the kernel, in order to emulate those ioctl()s )
Ah, I see. Thanks for the explaination.
---
Izumi Tsutsui
Module Name: src
Committed By: macallan
Date: Thu Feb 14 16:16:25 UTC 2013
Modified Files:
src/external/mit/xorg/server/xorg-server/hw/xfree86/xorgos: Makefile
Log Message:
set -DPCVT_SUPPORT on most hardware
Isn't it VGA specific?
---
Izumi Tsutsui
live-image targets for more ports.
(to build images without vnd(4) support)
---
Izumi Tsutsui
at this point.
Is this change reviewed by vm guys (especailly yamt@)?
If so it's much better to note it in the commit log.
---
Izumi Tsutsui
implementation which has many quirks.
As always communication could have been better.
Posting design description to public list is also better.
(especially for poor Tier II ports)
Thanks,
---
Izumi Tsutsui
---
Izumi Tsutsui
- FFS_NO_SNAPSHOT doesn't allow the creation of file system internal
snapshots. It was requested to reduce the size of INSTALL kernels only.
Hmm.
Then, is pseudo-device fss still useful with options FFS_NO_SNAPSHOT?
(most x68k has only 8~12MB)
---
Izumi Tsutsui
is safer,
I think.
(someone might claim we should add PRIxBUSADDRfoo, though)
---
Izumi Tsutsui
.
---
(though I'm not sure if it's msg.mbr file problem or not)
---
Izumi Tsutsui
/postinstall.
---
Izumi Tsutsui
ryo@ wrote:
I fixed evbarm/{ixdp425,nslu2} in the same way as matt@ did.
ok to commit?
Looks fine for me.
---
Izumi Tsutsui
I wrote:
chs@ wrote:
On Mon, Sep 17, 2012 at 03:40:12AM +0900, Izumi Tsutsui wrote:
chs@ wrote:
Module Name:src
Committed By: chs
Date: Tue Aug 14 15:46:21 UTC 2012
Modified Files:
src/sys/arch/evbarm/iq80310: iq80310_intr.c
chs@ wrote:
On Mon, Sep 17, 2012 at 03:40:12AM +0900, Izumi Tsutsui wrote:
chs@ wrote:
Module Name: src
Committed By: chs
Date: Tue Aug 14 15:46:21 UTC 2012
Modified Files:
src/sys/arch/evbarm/iq80310: iq80310_intr.c
Log Message:
move
build.sh will just work,
especially for -E case.
---
Izumi Tsutsui
dholland-sourcechanges@ wrote:
On Mon, Sep 17, 2012 at 03:26:02PM +0900, Izumi Tsutsui wrote:
Log Message:
Strengthen the entry about the openssl update, to clarify and emphasize
that just doing a non-update build isn't enough; you really do have to
nuke $DESTDIR.
Probably
after evcnt_init(),
rather than calling evcnt_attach_dynamic() in intr_establish() ?
It looks many other arm intr_init functions
(arm/ep93xx/ep93xx_intr.c, arm/xscale/ixp425_intr.c etc)
have the same problem, though.
(on the other hand marvell variants don't have evcnt at all)
---
Izumi
it might cause inode shortage because a number of nodes is
calculated before create_mfs_dev is called in makedev_main())
---
Izumi Tsutsui
images with device nodes
(-s: Generate mtree(8) specfile instead of creating devices)
and unlike fdesc compat pty nodes are created outside of /dev/pts,
so I'm afraid checking ptyfs is enough.
---
Izumi Tsutsui
martin@ wrote:
On Sat, Sep 08, 2012 at 01:34:37AM +0900, Izumi Tsutsui wrote:
Probably we should also remove ipty from MAKEDEV init target
and also add the above mount -t ptyfs ... line into
src/distrib/{amd64,i386}/cdroms/etc.rc and
src/distrib/sparc64/cdroms/installcd/etc.rc
because
?
It would be better to start it after we get some RC2 feedback?
---
Izumi Tsutsui
martin@ wrote:
On Sat, Sep 08, 2012 at 10:03:01PM +0900, Izumi Tsutsui wrote:
It would be better to start it after we get some RC2 feedback?
You mean for -current?
Yes, I meant the big revamp for all install media in -current after RC2.
We need to process #543 (or a variation
martin@ wrote:
On Fri, Sep 07, 2012 at 09:20:49PM +0900, Izumi Tsutsui wrote:
- What's the actual benefits on removing those device nodes on /dev?
Is it more important than possible fallouts in install materials?
Those nodes, if used together with ptyfs, create a serious security
risk
ipty in all_md and my freshly
installed NetBSD/i386 6.0_RC1 still has /dev/[pt]typ[01] nodes.
Should it be fixed even in netbsd-6?
(in that case we should also fix installimage script to use ptyfs though)
---
Izumi Tsutsui
.
Probably we should also remove ipty from MAKEDEV init target
and also add the above mount -t ptyfs ... line into
src/distrib/{amd64,i386}/cdroms/etc.rc and
src/distrib/sparc64/cdroms/installcd/etc.rc
because third party live CDs would use tmpfs based /dev
created by init(8) as installcd.
---
Izumi
martin@ wrote:
On Sat, Sep 08, 2012 at 01:34:37AM +0900, Izumi Tsutsui wrote:
Probably we should also remove ipty from MAKEDEV init target
and also add the above mount -t ptyfs ... line into
src/distrib/{amd64,i386}/cdroms/etc.rc and
src/distrib/sparc64/cdroms/installcd/etc.rc
because
901 - 1000 of 1202 matches
Mail list logo