ken [Tue, 27 Mar 2001 05:45:52 +0000 (05:45 +0000)]
Rewrite of the CAM error recovery code.
Some of the major changes include:
- The SCSI error handling portion of cam_periph_error() has
been broken out into a number of subfunctions to better
modularize the code that handles the hierarchy of SCSI errors.
As a result, the code is now much easier to read.
- String handling and error printing has been significantly
revamped. We now use sbufs to do string formatting instead
of using printfs (for the kernel) and snprintf/strncat (for
userland) as before.
There is a new catchall error printing routine,
cam_error_print() and its string-based counterpart,
cam_error_string() that allow the kernel and userland
applications to pass in a CCB and have errors printed out
properly, whether or not they're SCSI errors. Among other
things, this helped eliminate a fair amount of duplicate code
in camcontrol.
We now print out more information than before, including
the CAM status and SCSI status and the error recovery action
taken to remedy the problem.
- sbufs are now available in userland, via libsbuf. This
change was necessary since most of the error printing code
is shared between libcam and the kernel.
- A new transfer settings interface is included in this checkin.
This code is #ifdef'ed out, and is primarily intended to aid
discussion with HBA driver authors on the final form the
interface should take. There is example code in the ahc(4)
driver that implements the HBA driver side of the new
interface. The new transfer settings code won't be enabled
until we're ready to switch all HBA drivers over to the new
interface.
src/Makefile.inc1,
lib/Makefile: Add libsbuf. It must be built before libcam,
since libcam uses sbuf routines.
libcam/Makefile: libcam now depends on libsbuf.
libsbuf/Makefile: Add a makefile for libsbuf. This pulls in the
sbuf sources from sys/kern.
bsd.libnames.mk: Add LIBSBUF.
camcontrol/Makefile: Add -lsbuf. Since camcontrol is statically
linked, we can't depend on the dynamic linker
to pull in libsbuf.
camcontrol.c: Use cam_error_print() instead of checking for
CAM_SCSI_STATUS_ERROR on every failed CCB.
sbuf.9: Change the prototypes for sbuf_cat() and
sbuf_cpy() so that the source string is now a
const char *. This is more in line wth the
standard system string functions, and helps
eliminate warnings when dealing with a const
source buffer.
Fix a typo.
cam.c: Add description strings for the various CAM
error status values, as well as routines to
look up those strings.
Add new cam_error_string() and
cam_error_print() routines for userland and
the kernel.
cam.h: Add a new CAM flag, CAM_RETRY_SELTO.
Add enumerated types for the various options
available with cam_error_print() and
cam_error_string().
cam_ccb.h: Add new transfer negotiation structures/types.
Change inq_len in the ccb_getdev structure to
be "reserved". This field has never been
filled in, and will be removed when we next
bump the CAM version.
cam_debug.h: Fix typo.
cam_periph.c: Modularize cam_periph_error(). The SCSI error
handling part of cam_periph_error() is now
in camperiphscsistatuserror() and
camperiphscsisenseerror().
In cam_periph_lock(), increase the reference
count on the periph while we wait for our lock
attempt to succeed so that the periph won't go
away while we're sleeping.
cam_xpt.c: Add new transfer negotiation code. (ifdefed
out)
Add a new function, xpt_path_string(). This
is a string/sbuf analog to xpt_print_path().
scsi_all.c: Revamp string handing and error printing code.
We now use sbufs for much of the string
formatting code. More of that code is shared
between userland the kernel.
scsi_all.h: Get rid of SS_TURSTART, it wasn't terribly
useful in the first place.
Add a new error action, SS_REQSENSE. (Send a
request sense and then retry the command.)
This is useful when the controller hasn't
performed autosense for some reason.
Change the default actions around a bit.
scsi_cd.c,
scsi_da.c,
scsi_pt.c,
scsi_ses.c: SF_RETRY_SELTO -> CAM_RETRY_SELTO. Selection
timeouts shouldn't be covered by a sense flag.
simokawa [Tue, 27 Mar 2001 05:28:30 +0000 (05:28 +0000)]
Replace dyn_fin_lifetime with dyn_ack_lifetime for half-closed state.
Half-closed state could last long for some connections and fin_lifetime
(default 20sec) is too short for that.
obrien [Tue, 27 Mar 2001 01:34:58 +0000 (01:34 +0000)]
The common wisdom is to use the largest number of cylinders per group.
So bump the default from `16' to `22', which is the largest value allowed
with the current default block size. This change increases the the
group size from 32MB/g to 44MB/g on a 4GB SCSI disk.
rwatson [Mon, 26 Mar 2001 19:55:35 +0000 (19:55 +0000)]
o Update copyright date
o Revise description in light of commits over last month including:
- ACL editing library is now implemented
- ACLs are now implemented
ache [Mon, 26 Mar 2001 19:29:49 +0000 (19:29 +0000)]
rlines() checks:
1) really check for size overflow by checking negative value.
2) since mmap() not support files over INT_MAX size, add check for it
until either mmap() will be fixed or tail will be rewritted to handle
large files alternatively.
3) replace fseek(... file_size, SEEK_SET) with fseek(... 0L, SEEK_END)
to avoid off_t -> long cast
4) Use exit() if file is too big instead of warning and wrong logic
afterwards.
rwatson [Mon, 26 Mar 2001 17:53:19 +0000 (17:53 +0000)]
Introduce support for POSIX.1e ACLs on UFS-based file systems. This
implementation is still experimental, and while fairly broadly tested,
is not yet intended for production use. Support for POSIX.1e ACLs on
UFS will not be MFC'd to RELENG_4.
This implementation works by providing implementations of VOP_[GS]ETACL()
for FFS, as well as modifying the appropriate access control and file
creation routines. In this implementation, ACLs are backed into extended
attributes; the base ACL (owner, group, other) permissions remain in the
inode for performance and compatibility reasons, so only the extended and
default ACLs are placed in extended attributes. The logic for ACL
evaluation is provided by the fs-independent kern/kern_acl.c.
o Introduce UFS_ACL, a compile-time configuration option that enables
support for ACLs on FFS (and potentially other UFS-based file systems).
o Introduce ufs_getacl(), ufs_setacl(), ufs_aclcheck(), which
respectively get, set, and check the ACLs on the passed vnode.
o Introduce ufs_sync_acl_from_inode(), ufs_sync_inode_from_acl() to
maintain access control information between inode permissions and
extended attribute data.
o Modify ufs_access() to load a file access ACL and invoke
vaccess_acl_posix1e() if ACLs are available on the file system
o Modify ufs_mkdir() and ufs_makeinode() to associate ACLs with newly
created directories and files, inheriting from the parent directory's
default ACL.
o Enable these new vnode operations and conditionally compiled code
paths if UFS_ACL is defined.
A few notes:
o This implementation is fairly widely tested, but still should be
considered experimental.
o Currently, ACLs are not exported via NFS, instead, the summarizing
file mode/etc from the inode is. This results in conservative
protection behavior, similar to the behavior of ACL-nonaware programs
acting locally.
o It is possible that underlying binary data formats associated with
this implementation may change. Consumers of the implementation
should expect to find their local configuration obsoleted in the
next few months, resulting in possible loss of ACL data during an
upgrade.
o The extended attributes interface and implementation is still
undergoing modification to address portable interface concerns, as
well as performance.
o Many applications do not yet correctly handle ACLs. In general,
due to the POSIX.1e ACL model, behavior of ACL-unaware applications
will be conservative with respects to file protection; some caution
is recommended.
o Instructions for configuring and maintaining ACLs on UFS will be
committed in the near future; in the mean time it is possible to
reference the README included in the last UFS ACL distribution
placed in the TrustedBSD web site:
http://www.TrustedBSD.org/downloads/
Substantial debugging, hardware, travel, or connectivity support for this
project was provided by: BSDi, Safeport Network Services, and NAI Labs.
Significant coding contributions were made by Chris Faulhaber. Additional
support was provided by Brian Feldman, Thomas Moestl, and Ilmar Habibulin.
ru [Mon, 26 Mar 2001 14:53:33 +0000 (14:53 +0000)]
secure/ build fixes:
- TELNETOBJDIR is gone. `buildworld' already installs libtelnet.a
in ${WORLDTMP}/usr/lib, and we have LIBRARY_PATH pointing there.
- SSHDIR (formerly SSHSRC) is now shared between all SSH modules.
New LIBSSH is introduced for libssh.a (an internal static lib).
Previously, build without prior `obj' was broken; SSH modules
always looked for libssh.a in ${.OBJDIR}. Also, the dependancies
on the libssh.a were missing.
- libtelnet/ did not install the crypto version of telnet.h into
/usr/include/arpa.
- Removed BINOWN, BINMODE, BINDIR and SRCS with default values.
ru [Mon, 26 Mar 2001 12:49:05 +0000 (12:49 +0000)]
Do not build (and install) both secure/ and standard versions
of libtelnet, telnetd, and telnet. This only worked because
secure/ was listed late in SUBDIR in Makefile.inc1.
imp [Mon, 26 Mar 2001 08:05:20 +0000 (08:05 +0000)]
First step towards plugging the "pccard is trying to map into a BIOS
region for CIS reading" problem:
Use bus_alloc_resource to get the memory that we'll be using. Also
has the benefit of doing usage checking as well. This gets rid of the
ugly kludge that we had before for mapping pmem to vmem.
Second, move PIOCSRESOURCE to its own routine and make it conform more
to style(9) in the process.
alfred [Sun, 25 Mar 2001 23:32:55 +0000 (23:32 +0000)]
Don't call daemon() and setup our signal handlers until after we check
and do the unregister/reregister work.
Don't call syslog in the unregister/reregister code as we haven't called
openlog() yet.
Be a more conservative about accepting errno values from socket(2),
only EPROTONOSUPPORT means that the kernel isn't supporting it
something like INET6. The other possible errnos would be returned
if there was a mistake in the socket(2) call so remove them from the
list of "acceptable" return values.
brian [Sun, 25 Mar 2001 20:06:08 +0000 (20:06 +0000)]
Issue a ``quit'' after other ppp commands given on the command line and
then wait for the connection to be closed by the peer.
This means that commands such as ``pppctl ... show links'' will
display the correct output again (rather than truncating it depending
on how much data arrived in the last packet).
joerg [Sun, 25 Mar 2001 09:53:07 +0000 (09:53 +0000)]
This is another MFC candidate.
Fix a serious bug in sppp where anyone could obtain a successful PAP
authentication by supplying a null password. I've only stumpled across
the PR while browsing for all sppp-related PRs.
mikeh [Sun, 25 Mar 2001 04:57:05 +0000 (04:57 +0000)]
Merge various changes from OpenBSD and NetBSD.
o remove panic() in favor of err(3) and use err(3) functions
consistently throughout
o use stat(2)'s S_IS* macros rather than S_IF*
o [r]index -> str[r]chr
o convert some static buffers to dynamic ones
o use real tempfiles rather than reopening the same templates
o rename some functions that clash with libc
o convert wait_status from union to int and use wait(2) status macros
o fix multiple potential buffer overflows
o fix a few comments
o add $FreeBSD$
brian [Sun, 25 Mar 2001 01:34:27 +0000 (01:34 +0000)]
If we're adding an interface address that is already one of the interface's
aliases with the same netmask and destination, don't remove it and then
re-add exactly the same thing.
This means that static (non-sticky) routes that use the interface address
(or destination address) as a destination will not suddenly evaporate when
IPCP comes up (not unless the negotiated IPs have changed anyway).
bmilekic [Sat, 24 Mar 2001 23:47:52 +0000 (23:47 +0000)]
Move the atomic() mbstat.m_drops incrementing to the MGET(HDR) and
MCLGET macros in order to avoid incrementing the drop count twice.
Otherwise, in some cases, we may increment m_drops once in m_mballoc()
for example, and increment it again in m_mballoc_wait() if the
wait fails.
cg [Sat, 24 Mar 2001 23:10:29 +0000 (23:10 +0000)]
mega-commit.
this introduces a new buffering mechanism which results in dramatic
simplification of the channel manager.
as several structures have changed, we take the opportunity to move their
definitions into the source files where they are used, make them private and
de-typedef them.
the sound drivers are updated to use snd_setup_intr instead of
bus_setup_intr, and to comply with the de-typedefed structures.
the ac97, mixer and channel layers have been updated with finegrained
locking, as have some drivers- not all though. the rest will follow soon.
obrien [Sat, 24 Mar 2001 17:28:06 +0000 (17:28 +0000)]
Revert the warning elimination in rev 1.292. It breaks the Alpha build with:
/usr/src/usr.sbin/sysinstall/menus.c:1323: initializer element is not computable at load time
/usr/src/usr.sbin/sysinstall/menus.c:1323: (near initialization for `MenuNetworking.items[9].aux')
obrien [Sat, 24 Mar 2001 09:37:48 +0000 (09:37 +0000)]
List of distfile contents to delete using `rm' and `find -delete'.
Simular to the FREEBSD-Xlist used with `tar -X ... -xf'.
This file is typically used when one starts with the GDB anoncvs repo
rather than a release tarball.
obrien [Sat, 24 Mar 2001 09:01:31 +0000 (09:01 +0000)]
Sigh. I should know better than to commit a one line change from a PR
w/o a full compile test, even when the PR seemed so authoritative on the
subject...
obrien [Sat, 24 Mar 2001 08:44:28 +0000 (08:44 +0000)]
Restore the careful preservation of .depend (rev 1.78 of Makefile.i386)
that I removed in my last commit dealing with `make depend' bogons.
This commit has some races, but hopefully they are too short to matter.
Unfortuneatly, neither .newdep nor .olddep is removed by `make clean'.
obrien [Sat, 24 Mar 2001 08:27:57 +0000 (08:27 +0000)]
Fix a problem where we were switching npxproc from underneath processes
running in process context in order to run interrupt handlers. This
caused a big smashing of the stack on AMD K6, K5 and Intel Pentium (ie, P5)
processors because we are using npxproc as a flag to indicate whether
the state has been pushed onto the stack.
peter [Sat, 24 Mar 2001 04:40:49 +0000 (04:40 +0000)]
This is kind of a hack, but it should work. Currently, world is broken
because libc/rpc/key_call.c references uname(), and ps/print.c also
defines uname(), and ps is linked statically. This leads to a symbol
clash. The userland uname(3) kinda sucked anyway as the hostname
etc was too short. And since the libc rpc interface now uses
the utsname.nodename which gets truncated, I was tempted into doing
something about it. Create a new userland uname function, called
__xuname() which takes an extra argument that allows you to change
the size of the fields. uname() becomes a static inline function
in sys/utsname.h that passes the extra argument in. struct utsname
has its field members expanded by default now in userland.
We still provide a 'uname' externally linkable function for things
that either think that they ``know'' the utsname format and assume
32 character strings and bypass the include file, or objects that
are linked against old libcs. ie: just about every plausible
case that I can think of is covered. Should we ever change the
default lengths again, a libc major bump should not be required
as the size is now passed to the function.
XXX the uname(2) in the kernel is for FreeBSD 1.1 binary compatability!
All the uname(3) functions that are exported to userland are actually
implemented in libc with sysctl. uname(1) uses sysctl directly and
does not call uname(3).
obrien [Sat, 24 Mar 2001 01:58:31 +0000 (01:58 +0000)]
Import the setjump/longjump exception handling fixes from GCC 2.95.3.test3
that were removed from GCC 2.95.3.test4 and the subsequent release due
to problems on HP-UX. However, they work just fine on all the BSD's.
W/o these patches the following program segmentation faults if compiled
with -O2 (but not -Os or -O or -O0):
#include <stdio.h>
class A {
public:
A() { printf("c'tor A\n"); }
~A(){ printf("d'tor A\n"); }
};
class foo : public A {
public:
foo() { printf("C'tor foo\n"); throw 8; }
~foo() { printf("D'tor foo\n"); }
};
peter [Sat, 24 Mar 2001 00:34:12 +0000 (00:34 +0000)]
Slightly de-bogify the uname() declarations.
1: there is no global 'struct utsname utsname' in the kernel.
2: the __P() stuff handles the __STDC__ stuff, so the extra ifdef is
redundant.