roger [Fri, 1 Oct 1999 16:58:40 +0000 (16:58 +0000)]
Backout part of the changes made in 1.111
For unknown devices the output will now be
pci0: unknown card (vendor=0x109e, dev=0x0878) at 14.1 irq 19
instead of
pci0: unknown card DD^0878 (vendor=0x109e, dev=0x0878) at 14.1 irq 19
Before this change, the code used to take the PCI vendor id and translate it
into a three letter ASCII name.
For PnP devices, the vendor id _does_ map to a nice ASCII name
(eg Creative Labs PnP ID maps to "CTL", ESS PnP ID maps to "ESS")
But there is no such mapping for PCI devices, as can be seen by the
example above where the Brooktree PCI vendor ID maps to "DD^"
The PCI Special Interest Group confirmed they do not have any mappings
from vendor ID to ASCII.
peter [Fri, 1 Oct 1999 14:26:28 +0000 (14:26 +0000)]
Oops. That'll teach me to commit without testing. I either replaced
one trigraph with another, or completely missed the point. Kill it for
real this time.
peter [Fri, 1 Oct 1999 14:21:50 +0000 (14:21 +0000)]
Force the "calls" count for malloc types to be 64 bit where it keeps track
of the number of times a particular type has been used. It's rather easy
to overflow. One site I'm looking at seems to do it in a matter of days.
On the Alpha this is a no-op since 'long' is 64 bit already. The sole
user of this interface seems to be vmstat -m and friends which will need
a recompile. The overheads of using a 64 bit int should be pretty light
as the kernel just does "calls++" type operations and that's it.
Let sleep(1) handle fractions of a second (up to nanosecond).
This is a conservative change. It does the same thing in weird
cases like the old one. For example, 'sleep abcd' still sleeps
for zero seconds. `sleep 10.a' and `sleep 10.05aa' do the best
and not abort (ie: 10.a == 10 seconds, 10.05a == 10.05 seconds).
mdodd [Fri, 1 Oct 1999 05:24:27 +0000 (05:24 +0000)]
Turn off and remove the 'old' if_ep ISA/EISA/PCCARD driver.
Turn on the 'new' if_ep driver which supports:
ISA 3c509
MCA 3c529
EISA 3c579
PCCARD 3c589
I think all we're missing is support for the VME bus and S-100 bus
Etherlink III cards.
The new code has been tested by a number of people and all the important
bits work. I've not been able to test the EISA code but will do so once
my hardware arrives. Since I've changed nothing in the EISA code I suspect
it will perform the same manner as before.
Future changes involve whacking the ISA and PCCARD front ends to use
newbus and to convert the driver to bus_space and make it use ifmedia.
This is the first working network driver that supports MCA bus devices btw.
This fix allows cards inserted in one of the bottem-most slots on an xp1000
(behind the built-in ppb on hose 1) to be found:
When testing the adaptec controller on alpha, I realized I misread the xp1000
documentation and the way I'm calculating the bus number for PCI
config space accesses on the tsunami is wrong. I had thought that a bus
behind a ppb should be numbered as the nth bus in that hose, but it
actually needs to be the nth global bus within the system. The bus number
for the primary bus on a hose must always remain 0 when calculating config
space addresses.
Make the fxp driver work on alpha, rather than panic the machine on boot
and/or when using the card.
o Convert the driver to using bus_space. This allows alphas with
fxp's to boot, rather than panic'ing because rman_get_virtual()
doesn't really return a virtual address on alphas.
o Fix an alpha unaligned access error caused by some misfeature of
gcc/egcs: if link_addr & rbd_addr in the fxp_rfa struct are 32 bit
quantities, egcs will assume they are naturally aligned. So it will do
a ldl & some shifty/masky to twiddle 16 bit values in fxp_lwcopy().
However, if they are 16-bit aligned, the ldl will actually be done on
a 16-bit aligned value & we will panic with an unaligned access
error... Changing their definition to an array of chars seems to fix
this. I obtained this from NetBSD.
Add ahc driver. The adaptec driver works just dandy on my xp1000 using
a 2940UW. The dp264 (and ds20, I think) have an AIC7895 on board so it
is important the ahc driver be in GENERIC so that FreeBSD can install on
these boxes.
Remove the D_NOCLUSTER[RW] options which were added because vn had
problems. Now that Matt has fixed vn, this can go. The vn driver
should have used d_maxio (now si_iosize_max) anyway.
Test the slices for openness before we close them; doing it the other way
around meant that the higher level close routine never gets called.
(phk is on the road; this is a quick fix to get things working and may need
more polish)
Careless use of struct proc *p caused major problems. 'p' is allowed to
be NULL in this function (nfs_sigintr). Reorder the statements and guard
them all with a single if (p != NULL).
doscmd heavily depends on struct sigcontext which luckily is mostly passed
between functions as usion regcontext_t. By redefining union regcontext_t in
terms of mcontext_t almost all bases are covered.
It also seems to me that doscmd was in a transitional state. The redundant
definitions made it difficult to get a clear overview and could easily cause
oversight. To make sure my changes were ok, I went as far as to complete the
transition. It was not exactly necessary, but I expect to have to come back
here some more ("whistle" if I'm wrong :-).
Make FreeBSD less conservative in determining when to return a cookie
error for a directory. I have made this change after a great deal of
review although I cannot be absolutely sure that this meets the spec.
The issue devolves into whether changes in an underlying (UFS) directory
can cause NFS directory blocks to be renumbered. My read of the code
indicates that NFS directory blocks will not be renumbered, which means
that the cookies should still remain valid after a change is made to
the underlying directory. This being the case, a cookie error should
not be returned when a change is made to the underlying directory and,
instead, the NFS client should rely on mtime detection to invalidate and
reload the directory.
The use of mtime is problematic in of itself, due to insufficient
resolution, which is why I believe the original conservative error
handling was done. Still, there have been dozens of bug reports by
people needing solaris<->FreeBSD interoperability and these have to
be accomodated.
sigset_t change (part 5 of 5)
-----------------------------
Most of the userland changes are in libc. For both the alpha
and the i386 setjmp has been changed to accomodate for the
new sigset_t. Internally, libc is mostly rewritten to use the
new syscalls. The exception is in compat-43/sigcompat.c
The POSIX thread library has also been rewritten to use the
new sigset_t. Except, that it currently only handles NSIG
signals instead of the maximum _SIG_MAXSIG. This should not
be a problem because current applications don't use any
signals higher than NSIG.
There are version bumps for the following libraries:
libdialog
libreadline
libc
libc_r
libedit
libftpio
libss
These libraries either a) have one of the modified structures
visible in the interface, or b) use sigset_t internally and
may cause breakage if new binaries are used against libraries
that don't have the sigset_t change. This not an immediate
issue, but will be as soon as applications start using the
new range to its fullest.
NOTE: libncurses already had an version bump and has not been
given one now.
NOTE: doscmd is a real casualty and has been disconnected for
the moment. Reconnection will eventually happen after
doscmd has been fixed. I'm aware that being the last one
to touch it, I'm automaticly promoted to being maintainer.
According to good taste this means that I will receive a
badge which either will be glued or mechanically stapled,
drilled or otherwise violently forced onto me :-)
NOTE: pcvt/vttest cannot be compiled with -traditional. The
change cause sys/types to be included along the way which
contains the const and volatile modifiers. I don't consider
this a solution, but more a workaround.
sigset_t change (part 4 of 5)
-----------------------------
The compatibility code and/or emulators have been updated:
iBCS2 now mostly uses the older syscalls. SVR4 now properly
handles all signals. This has been achieved by using the
new sigset_t throughout the emulator. The Linuxulator has
been severely updated. Internally the new Linux sigset_t is
made the default. These are then mapped to and from the
new FreeBSD sigset_t.
Also, rt_sigsuspend has been implemented in the Linuxulator.
Implementing this syscall basicly caused all this sigset_t
changing in the first place and the syscall has been used
throughout the change as a means for testing. It basicly is
too much work to undo the implementation so that it can
later be added again.
A special note on the use of sv_sigtbl and sv_sigsize in
struct sysentvec:
Every signal larger than sv_sigsize is not translated and is
passed on to the signal handler unmodified. Signals in the
range 1 upto and including sv_sigsize are translated.
The rationale is that only the system defined signals need to
be translated.
The emulators also have been updated so that the translation
tables are only indexed for valid (system defined) signals.
This change also fixes the translation bug already in the
SVR4 emulator.
sigset_t change (part 3 of 5)
-----------------------------
By introducing a new sigframe so that the signal handler operates
on the new siginfo_t and on ucontext_t instead of sigcontext, we
now need two version of sendsig and sigreturn.
A flag in struct proc determines whether the process expects an
old sigframe or a new sigframe. The signal trampoline handles
which sigreturn to call. It does this by testing for a magic
cookie in the frame.
The alpha uses osigreturn to implement longjmp. This means that
osigreturn is not only used for compatibility with existing
binaries. To handle the new sigset_t, setjmp saves it in
sc_reserved (see NOTE).
the struct sigframe has been moved from frame.h to sigframe.h
to handle the complex header dependencies that was caused by
the new sigframe.
NOTE: For the i386, the size of jmp_buf has been increased to hold
the new sigset_t. On the alpha this has been prevented by
using sc_reserved in sigcontext.
sigset_t change (part 2 of 5)
-----------------------------
The core of the signalling code has been rewritten to operate
on the new sigset_t. No methodological changes have been made.
Most references to a sigset_t object are through macros (see
signalvar.h) to create a level of abstraction and to provide
a basis for further improvements.
The NSIG constant has not been changed to reflect the maximum
number of signals possible. The reason is that it breaks
programs (especially shells) which assume that all signals
have a non-null name in sys_signame. See src/bin/sh/trap.c
for an example. Instead _SIG_MAXSIG has been introduced to
hold the maximum signal possible with the new sigset_t.
struct sigprop has been moved from signalvar.h to kern_sig.c
because a) it is only used there, and b) access must be done
though function sigprop(). The latter because the table doesn't
holds properties for all signals, but only for the first NSIG
signals.
signal.h has been reorganized to make reading easier and to
add the new and/or modified structures. The "old" structures
are moved to signalvar.h to prevent namespace polution.
Especially the coda filesystem suffers from the change, because
it contained lines like (p->p_sigmask == SIGIO), which is easy
to do for integral types, but not for compound types.
NOTE: kdump (and port linux_kdump) must be recompiled.
Thanks to Garrett Wollman and Daniel Eischen for pressing the
importance of changing sigreturn as well.
sigset_t change (part 1 of 5)
-----------------------------
Rename sigaction, sigprocmask, sigpending and sigsuspend to
osigaction, osigprocmask, osigpending and osigsuspend (resp)
and add new syscalls for them to support the new sisgset_t
without breaking existing binaries.
Change the prototype of sigaltstack to use the typedef stack_t
instead of struct sigaltstack to reflect that it is SUSv2
compliant.
Also, rename sigreturn to osigreturn and add a new syscall
to support the modified stackframe. The change is caused by
sigreturn operating on ucontext_t now and the fact that
siginfo_t has been updated to conform to SUSv2.
vinum_lpi: Add information about progress of rebuildparity and
checkparity commands.
listconfig: Print object counts in a different format, don't refer to
the internal number of slots allocated:
Caused-confusion-to: Norbert Meissner <norbert.meissner@daimlerchrysler.com>
vinum_info: Change format of device numbers so that they fit in the
column (ignore high-order minor number bits, which aren't really of
much interest).
struct plex: Add variables checkblock and rebuildblock to keep track
of parity check and rebuild operations. This enables us
to stop the operation and restart at a later time.
enum parityop: Trivial enum to decide what parityops() is going to do.
open_drive: Don't call set_drive_state to take a drive down, do it
ourselves. This breaks a vicious circle which caused
vinum to dereference a null vp if device nodes were
missing.
get_empty_volume: initialize plexes to -1 (not allocated)
remove_drive_entry:
Remove recurse parameter (there's nothing below a drive in the hierarchy).
Use remove_sd_entry to remove sds, don't do it ourselves.
Log errors, don't throw rude remarks.
remove_plex_entry:
Don't use plex->subdisks as a loop limit, it gets changed in the
loop. This caused some removals to only remove half the subdisks.
Change logging of some "impossible" situations.
remove_volume_entry:
Use remove_plex_entry to remove plexes, don't do it ourselves.
update_sd_config:
Use set_sd_state to do the work.
* add a non-reset device- will not reset the channel on open. you
will have to mknod yourself for now.
* don't eat the first write()
* partial rvplayer fix- don't panic on unaligned writes unless our
feeder chain requires them for downconversion. a fuller fix is
on the way.
Use rmand_get_bus{tag,handle} rather than hard wiring things to
I386_BUS_SPACE_IO. Compiles now on the Alpha, but likely will not
work due to bus space address <-> virtual address mapping bogons that
work for i386 but not alpha.
peter [Tue, 28 Sep 1999 16:58:46 +0000 (16:58 +0000)]
Back out my backout, it was already posix compliant. Any new fields are
required to be "announced" by a new bit in sa_flags to indicate the
program is aware of and has taken care of them. eg: SA_SIGINFO means
the program has used the sa_siginfo field (versus sa_handler).
peter [Tue, 28 Sep 1999 15:40:17 +0000 (15:40 +0000)]
Fix previous commit. The standards specifically say: "The structure
sigaction, used to describe an action to be taken, is defined in the
header <signal.h> to include at least the following members:"
^^^^^^^^
A sigaction defined on stack with essentially random contents may have
just about anything underneath fields that the program doesn't know about.
It is not safe to delete the bzero.
Explicitly use sigemptyset to clear a sigset_t. Explicit
initialization of sa_flags added so that we can lose the bzero.
IIRC, this code is not used anymore since the addition of
ncurses. Commit the change anyway so, just to be safe.
Explicitly use sigemptyset to clear a sigset_t. Explicit
initialization of sa_flags added so that the 'struct sigaction'
can be declared local in both functions that use the global
(static) declaration. Remove the global declaration.
In sbflush(), don't exit the while loop too early: this can cause
an empty mbuf to stay in the queue, then causing a needless panic
because sb_cc == 0 and sb_mbcnt != 0.
But we still need to panic rather than endlessly looping if, for
some reason, sb_cc == 0 and there are non-empty mbufs in the queue.
Make sure file after VOP_OPEN is VMIO'd when transfering control from
a lower layer to an upper layer. I'm not sure how necessary this is
for reading.
Fix bug in union_lookup() (note: there are probably still several bugs
in union_lookup()). This one set lerror as a side effect without
setting lowervp, causing copyup code further on down to crash on a null
lowervp pointer. Changed the side effect to use a temporary variable
instead.