uvmfault_anonget: clean-up, improve some comments, misc.

(rmind)

From Aleksey Cheusov: Don't make it easy for compromised systems to bypass
ASLR protections by providing the mapping addresses of programs to everyone.

(christos)

Note some long-term difficult goals.

(jruoho)

pmap_map_ptes: use cpu_load_pmap() to handle i386 PAE case.
Spotted by cherry@

(rmind)

Remove a noisy debugging printf that triggers on bigendian platforms.
(Reported upstream as well)

(riz)

Document additional settings for onpanic, from options(4).

(riz)

Attempt to improve the documentation of ddb.onpanic, from options(4).

(riz)

Catchup with rmind-uvmplock merge.

(cherry)

No need to repeat checking want_resched then call preempt() in
T_AST case, so check it only once.

(nakayama)

Remove noneed platid_match() in consinit().

(nonaka)

Delete unused define.

(nonaka)

fix some more paths for GCC 4.5.

(mrg)

Added __packed define to be able to use eMbedded Visual C++ for building hpcboot.exe.

(nonaka)

make this work with mknative-gcc:
- libgcc.mvars only depends upon config.status/Makefile, not any built thing
- copy the unwind.h creation rule into a separate target, just for mknative-gcc

(mrg)

make this build on amd64.

XXX: need to confirm that all the objects are built but this is
XXX: enough to make GCC happy.

(mrg)

don't try to declare alloca() (wrongly) on netbsd.

(mrg)

cast a uint8_t * to a char * for a function that takes char *.

(mrg)

output stuff to the right libiberty dir for gcc 4.5.

(mrg)

fix a bunch of paths to look in the right places for gcc 4.5.

(mrg)

Don't check DISK_INFO_FLAG_CHS_UNCERTAIN.
Because DISK_INFO of SD on windows mobile 5 is reported it.

(nonaka)

Reduced the use of the magic number.

(nonaka)

Fix bit name ISR_UB.  Not _UE.
And add comments for bit names from datasheet of PXA255.

(kiyohara)

set PXA2X0_I2C_BASE to sc->sc_addr in the correct position.

(nonaka)

Make sure to include "opt_multiprocessor.h".
Add missing ). (spotted by uebayashi@)

(matt)

Fix bug pointed out by njoly@.

(jruoho)

Add pmap_tlb_asid_check and use it verify asid handling.
The shared pmap is now in a state that mips could switch to it.

(matt)

Sort sections, fix typos.

(wiz)

Fix typos, sort sections.

(wiz)

Use new names for the pmap_syncicache routines

(matt)

Separate code out from pmap_tlb.c which handles the synchronization of the
icache across CPUs.

(matt)

Add pmap_md_tlb_info_attach prototype.

(matt)

get this some what working with GCC 4.5.  i at least have managed to
get crtstuff and libgcc to build for amd64, and have linked and tested
/bin/sh with all these GCC 4.5 compiled part.

(mrg)

Cleanup comments in pmap_tlb.c.  Attach tlb evcnts.
eliminate ti_mask
Cleanup some of the MP code.  Conditionalize shootdown code.

(matt)

fix the file list and make this work.

(mrg)

fix this to actually build and seem to work.

(mrg)

Move inclusion of sys/endian.h from cd9660_eltorito.c to cd9660.h
since actual user is not cd9660_eltorito.c but iso.h and
cd9660_eltorito.h.

Actually, include order/place of sys/endian.h doesn't matter on
netbsd since it is always included by sys/types.h but it's not
true on other system.  This should fix cross build breakage on
freebsd introduced by rev. 1.16 of cd9660_eltorito.c.
Problem reported and fix suggested on twitter.

(enami)

Move some MD parts back to the booke pmap.c.  Cleanup initialization a bit.

(matt)

Switch to using the common <common/pmap/tlb/tlb.h>

(matt)

Add GPIO_PIN_INTR and GPIO_PIN_INTR_HIGH (placeholders for now).

(matt)

Initial work on a NOR driver; it's commited here as is,
so I can cooperate with Cliff Neighbors on finishing it.

Not any way connected to the build.

(ahoka)

After each strsep(&p) check that p is valid.
If not the .meta file is corrupted (build interrupted?)
issue a warning and declare target oodate.

(sjg)

Add a small summary page for altq. From the list in PR # kern/21345.

(jruoho)

Add a small summary page for sysmon. From the list in PR # kern/21345.

(jruoho)

Regen.

(jruoho)

Add SMO8800.

(jruoho)

make LSE prepend the rest of the shims in they exist

(kefren)

Provide the pci_* routines for modules to use.

(matt)

ibm4xx now uses <powerpc/pci_machdep.h>

(matt)

Add support pci_intr_setattr.
Export non-inline version of pci api for modules (_MODULE is defined).
Fix definition of pc_conf_hook and pc_conf_interrupt.
Switch to using inlines instead of macros.
Switch ibm4xx to use <powerpc/pci_machdep.h>

(matt)

Add inline accessors for pba_pc and pa_pc so that <machine/pci_machdep.h>
can use them in inline functions.

(matt)

bump up SYMTAB_SPACE

(matt)

Add member sc_addr to struct pxa2x0_i2c_softc.
And use sc_addr to bus_space_map() instead of macro-directly.

(kiyohara)

Include liblzma to unbreak gzip on the install CD. (Hi Joerg!)

(martin)

Add small comment.

(jruoho)

Get rid of RUN_ONCE(9). Should fix PR # kern/44043.

(jruoho)

Contrary to the previous, use aprint_error(9) but suppress warnings
caused by the (optional) BIOS-based dynamic voltage and frequency scaling.

(jruoho)

don't bother casting the return value of malloc.

(mrg)

use offsetof() instead of some strange local hack.

(mrg)

there's no fr_arg[6] or fr_arg[7] -- they're called fr_fp and fr_pc.
use the right names and avoid some array bounds violations picked up
by GCC 4.5.  it's a real bug, but the code would have been generated
correctly for the sparc* ABIs.

(mrg)

apply some -Wno-error/-Wno-enum-compare/-Wno-array-bounds.

this should be looked at more closely.

(mrg)

when checking for SKSV data, use the same test as before.  fixes an
always true conditional GCC 4.5 found, and is also a real bug fix.

(mrg)

fix the paths for GCC 4.5.  don't build the non-tools gmp/mpfr/mpc libs
yet since we don't have them fully working.  XXX.

(mrg)

port this change from our gcc 4.1 tree.  not entirely sure it's correct
yet, but not yet able to test ...

revision 1.5
date: 2011/04/10 16:50:18;  author: joerg;  state: Exp;  lines: +2 -2
Reduce difference between src/tools/gcc and src/gnu/usr.bin/gcc4
configuration. All but the target to helper programs should be the same.
Mark include directories as sysroot-relative.

(mrg)

avoid installing limits.h and friends.

(mrg)

fix the parts that build libgcc/crtstuff, etc., for GCC 4.5.

(mrg)

This is no longer needed.

(matt)

Make pmppc use stock <powerpc/pci_machdep.h>
Move some private declartions to <pmppc/dev/mainbus.h>

(matt)

don't compile tag functions if SMALL is defined. Should fix the build

(kefren)

remove some always true if () checks GCC 4.5.3 picks up.

(mrg)

fix the length argument to a memcpy() to refer to the size of the
destination, not the size of the pointer.  real bug on 64 bit platforms.

(mrg)

fix an off by one array overflow found by GCC 4.5.3.

(mrg)

fix an operator precedence error picked up by GCC 4.5.3.  real bug.

(mrg)

deal with __builtin_va_start()'s introduction.

(mrg)

fix a potential off-by-two array overflow found by GCC 4.5.3.

(mrg)

use %option noinput since we need it.  found by GCC 4.5.3.

(mrg)

remove an always true test found by GCC 4.5.3.

(mrg)

initialise from_cs[1] instead of from_cs[0] twice.  from GCC 4.5.3.

(mrg)

don't write "(u_char) NULL".

(mrg)

add a missing part from rev 1.10 and s/newline/newlinex/.  found by GCC 4.5.3.

(mrg)

apply some -Wno-error and/or -fno-strict-aliasing.

all of this should be looked at closer, but some of them are not
very trivial.

(mrg)

Take the sizeof the *correct* structure.  Now iscsid returns sensible
data to iscsictl that the display code can display (more) sensibly.

(riz)

Actually used SSE branch. Noticed by enami@

(joerg)

Include liblzma to unbreak gzip on the install CD.

(joerg)

New sentence, new line.

(wiz)

add a sentence regarding multiple tags

(kefren)

print multiple tags if exists, comma separated

(kefren)

backout previous.

(nonaka)

Make it compile when PIC_DEBUG is defined.

(rjs)

The referred-to comment was moved to vfs_vnode.c; update the pointer.

(riz)

use pmf_event_inject() instead of call lcd_blank() directly.

(nonaka)

Fix a typo. Markup nit. Remove trailing whitespace, and uncustomary dot at end of SEE ALSO.

(wiz)

Remove trailing whitespace.

(wiz)

Add bicons to constab.

(kiyohara)

Split pxa2x0_hpc/wzero3.

(kiyohara)

More stride for PSIONTEKLOGIX NETBOOK PRO.

(kiyohara)

Initial import of hashtree functions into othersrc.

The hashtree, or Merkle tree algorithms, split a file up into blocks,
and calculate a message digest on those blocks.  Subsequent digests
are concatenated onto the previous digests.  If there is more than one
digest calculated, a higher level of digest is then calculated across
the lower-order digests, to provide a tree-like structure of message
digests.

If there is only one digest calculated, then the hashtree in this
degenerate case is the same as the digets of the file.

% hashtree -n 1 dist/hashtree.c
HASHTREE/sha1/1/10358/10358 (dist/hashtree.c) = 253f185d39783fd29ca6a7936fee4b50e3775f39
% sha1 dist/hashtree.c
SHA1 (dist/hashtree.c) = 253f185d39783fd29ca6a7936fee4b50e3775f39
%

hashtree(1) can be used to specify the hashing algorithm used
(including hmac algorithms), and the size of the blocks.

% hashtree -b 4096 dist/hashtree.c
HASHTREE/sha1/2/10358/4096 (dist/hashtree.c) = 1a958db53705b1ef4ef4e7418f7aa6e25226c4468783684b4ff3d152a66a38e7ad3e5c8d88ad0851b5f916e7d015015c6e0ddf0cef4cf7af2d48a0ad6db45855c61ea5454b579221bcbeeb5e282e626d
%

Alternatively, the number of desired blocks can be specified, and
hashtree will work out the block size from that parameter and the size
of the file.  By default, a blocksize of 1024 is used.  There is the
possibility of producing very large digests using hashtree(1):

% hashtree dist/hashtree.c
HASHTREE/sha1/2/10358/1024 (dist/hashtree.c) = 6a765d5809534fa75fd2f4f2f88168e65cb1182f91ac14db77b0052ae083743c10ed0ccea98ebf44a945eb72ee0303a8b5b89a2a590ec2e684418942bae3e0be34567bd12721167239fdf641c24ec51b59426924a32760adf11370aca375d60a97364613f93e1e44f848a3557c141b7422aa7e63fef99ff4e056124513c228c9a33e0dff08e6baa2903f959f428dd60f9a6390b0d4108d0826713d33ab6e5520fe20e61ac700659eb5e6a1cf5653da05d65c0d792cd3affaadd037372cfbf7085e14caea50dd89e2790cd982824fec985e2a97fcbbc415e91c27dc4f3089ec446784262269677aa1736fba2e5a768114

Merkle trees can be used to implement Lamport's signature scheme - see

http://www.rsa.com/rsalabs/node.asp?id=2003

and there are also uses in ZFS:

http://blogs.oracle.com/bonwick/entry/zfs_end_to_end_data

(although indirect pointers can be a tad difficult here?)

Hash trees are written up more generally in:

http://en.wikipedia.org/wiki/Hash_tree

There are also uses of the tiger algorithm being used to produce more
specific "tiger trees".

Status:

Vendor Tag: CROOKS
Release Tags: hashtree-base

(agc)

Alloc dynamically sockunions in sou.
Teach route(8) about multiple tags. E.G.: -tag 100,20,33
XXX: needs documentation

(kefren)

learn mpls interface how to prepend multiple shims by using a vector of
smpls_addrs in sockaddr_mpls. The number of smpls_addrs is found from
smpls_len. First label encountered is BoS.
XXX: need to do the same for LSE and this feature needs to be documented.

(kefren)

Use kthread_join() to wait for the termination of the helper thread.

(hannken)

Add a few explicit casts for sign mismatches.

(joerg)

When reading or writing the backing store us POSIX_FADV_NOREUSE access hint
to prevent read ahead.

(hannken)

Make sure sensor state is valid before calling sysmon_envsys_sensor_attach().

Ok: Paul Goyette <pgoyette@netbsd.org>

(hannken)

Revert rev 1.42 which is not necessary after rev 1.41.

(tsutsui)

fix typo.

(nonaka)

Use CFATTACH_DECL_NEW().

(kiyohara)

Revert the previous commit as it was wrong/redundant.

(jruoho)

Sort option descriptions. Add -v to SYNOPSIS. Remove empty EXAMPLES section.
Increase Bl width so everything fits.

(wiz)

Sort option descriptions. Add matpax(1) to SEE ALSO. Increase Bl width so everything fits.

(wiz)

Fix typo in option letter.

(wiz)

use --sysroot with GCC 4.5. (maybe could with 4.1 too?  the 4.1 code
here is broken right now, anyway, ever since the sysroot changes were
merged into our tree.)

(mrg)

avoid code that looks like this:
x = x = y;
since GCC 4.5 complains about it having sequence point issues.

(mrg)

Sort sections.

(wiz)

add a missing " to a string literal in some #if'd out code that GCC 4.5 found.

(mrg)

this needs %option noinput.  thanks to martin.

(mrg)

pull across these revisions from gcc 4.1:

configure:
revision 1.12
date: 2011/03/10 15:14:49;  author: skrll;  state: Exp;  lines: +1 -1
Look for TLS on hppa on netbsd.

configure.ac:
revision 1.11
date: 2011/03/10 15:14:49;  author: skrll;  state: Exp;  lines: +1 -1
Look for TLS on hppa on netbsd.

(mrg)

pull across these revisions from gcc 4.1:

revision 1.8
date: 2011/06/11 00:57:29;  author: matt;  state: Exp;  lines: +2 -0
If we are compiling -msoft-float, define _SOFT_FLOAT

revision 1.7
date: 2011/03/12 07:57:38;  author: matt;  state: Exp;  lines: +3 -1
Define HAVE_AS_TLS until our build tools figure out how to do it.

revision 1.6
date: 2011/02/25 22:36:10;  author: matt;  state: Exp;  lines: +7 -0
Explicitly make sure TARGET_SECURE_PLT is defined correctly rather than
relying on HAVE_AS_REL16 from "auto-host.h".

revision 1.5
date: 2011/02/10 07:18:29;  author: matt;  state: Exp;  lines: +3 -0
Default NetBSD to -msecure-plt now.

(mrg)

pull across these revisions from gcc 4.1:

revision 1.21
date: 2011/01/23 06:29:52;  author: matt;  state: Exp;  lines: +1 -1
Default mips64 to softfloat.

revision 1.22
date: 2011/02/10 07:18:29;  author: matt;  state: Exp;  lines: +3 -0
Default NetBSD to -msecure-plt now.

(mrg)

#include <sys/proc.h> before using struct proc.p_timerpend

(cherry)

pull across these revisions from gcc 4.1:

revision 1.9
date: 2011/04/15 13:42:22;  author: joerg;  state: Exp;  lines: +2 -1
Move the headers for compiler and platform specific intrinsincs from
/usr/include to /usr/include/gcc-4.1.

revision 1.8
date: 2011/04/10 21:25:56;  author: joerg;  state: Exp;  lines: +1 -0
Add back a stylish new line.

revision 1.7
date: 2011/04/10 16:50:18;  author: joerg;  state: Exp;  lines: +20 -12
Reduce difference between src/tools/gcc and src/gnu/usr.bin/gcc4
configuration. All but the target to helper programs should be the same.
Mark include directories as sysroot-relative.

adjust the paths to gcc-4.5, and add an XXX note about this ugliness.

(mrg)

Add #define __PMAP_PRIVATE

(matt)

Switch to use new common cpu_ast routine.

(matt)

Add a common cpu_ast(l, ci) routine.
Add a ci_pmap_asid_cur for BOOKE debugging.

(matt)

Add DEBUG around a variable.

(matt)

In e500_spl0, grab ci after disabling interrupts.
Consolidate a KASSERT.

(matt)

Adjust the #ifdefs such that ccd(4) and cgd(4) show in
modstat(8) even if built into the kernel.

(jruoho)

pull across from gcc 4.1:

revision 1.2
date: 2011/02/05 00:37:40;  author: joerg;  state: Exp;  lines: +2 -2
Explicitly use template to allow building with the more strict
template lookup in clang. From FreeBSD.

(mrg)

pull across from gcc 4.1:

revision 1.2
date: 2011/05/25 15:51:55;  author: joerg;  state: Exp;  lines: +2 -2
Don't define explicit instantiation as inline.

(mrg)

move the gcc 4.5 usr.bin build makefiles to the right place.

(mrg)

*sigh* move these files into the directories i meant for them to exist in.
start actually porting to gcc 4.5 in the process.

(mrg)

remove stuff from the wrong directory.  oops.

(mrg)

Nuke the mfsprg2 %r3 after sctrapexit.  curlwp is already in %r13 and the
value fetched to %r3 isn't used.

(matt)

There is no .../5.99 path to use for PKG_PATH, so change it to 5.1 instead.
Also refer the reader to pkg_install.conf as an alternative.

(erh)

Remove #if 0 block

(matt)

remove a #if 0 block

(matt)

By default, enable wmihp(4) instead of hpqlb(4), as it has a better chance
of working with a greater number of HP laptops. Ok cegger@.

(jruoho)

Add a common struct containing tlb ops needed by the tlb pmap.c

(matt)

front end to build gcc 4.5 components

(mrg)

prepare to build native gcc 4.5, don't build native gcc 4.1, for HAVE_GCC=45

(mrg)

build stuff for mpc, mpfr and gmp.  mpfr is mine, the others are from chuq.

XXX: i don't think gmp works yet.  it wants multiple files with the same
basename but in different subdirs to link into one library...

(mrg)

Add an extra assert to verify we aren't trying to process softints while
interrupts are being serviced.

(matt)

Switch to using dcache_wbinv_page

(matt)

Add a separate "audit" comment to the tar and pax frontends for mat, which
compare each component of the archive with its entry in the file system.
Sizes, all digests, oweners, group owners, mtimes and symbolic link targets
are all compared.

Add a test for the new audit command:

cd d && env LD_LIBRARY_PATH=${LIBMATDIR} ../${PROG} avvf ../archive2.mat > archive2.audit
cd d && env LD_LIBRARY_PATH=${LIBMATDIR} ../${PROG} tvvf ../archive2.mat > archive2.list
-diff d/archive2.list d/archive2.audit

(from the testing target in the mat Makefile).

Update the matpax(1) manual page, per wiz, to describe the pax arguments,
rather than the tar arguments.

(agc)

Reap the ci_ev_soft* evcnts since every variant implements __HAVE_FAST_SOFTINTS

(matt)

forward declare cpu_info and trapframe.

(matt)

add a NetBSD_DISABLED rule for cfns.h.

(mrg)

initial port to HAVE_GCC = 45.  mknative-gcc parts are from me
and are largely untested, most of the rest is from chuq.

(mrg)

add HAVE_GCC=45 support.

(mrg)

add some "NetBSD_Disabled*" rules to avoid r/o src tree build issues.

(mrg)

Rename acpi_get_node() to acpi_match_node() for consistency.

(jruoho)

Note the potential danger in the options. Fixes PR lib/31112.

(jruoho)

import the non-arch specific makefiles we use to build GCC 4.1
as a base reference.

Status:

Vendor Tag: netbsd
Release Tags: netbsd-20110620

(mrg)