overflows.
It sure would be nice if there was an unmapped page between the PCB and
the stack (and that the size of the stack was configurable!). With the
way things are now, the PCB will get clobbered before the double fault
handler gets control, making somewhat of a mess of things. Despite this,
it is still fairly easy to poke around in the overflowed stack to figure
out the cause.
Unstaticize a function in scsi/scsi_base that was used, with an undocumented
option.
My last count on the LINT kernel shows:
Total symbols: 3647
unref symbols: 463
undef symbols: 4
1 ref symbols: 1751
2 ref symbols: 485
Approaching the pain threshold now.
*' instead of caddr_t and it isn't optional (it never was). Most of the
netipx (and netns) pr_ctlinput functions abuse the second arg instead of
using the third arg but fixing this is beyond the scope of this round
of changes.
prototypes don't go missing again. Also added -Winline so that some
doubtful (non-)inlines get fixed.
bsd.kmod.mk:
Also added `-Wreturn-type -Wimplicit -Wnested-externs' to catch up
with the kernel.
#includes to get prototypes.
pci now uses a different interrupt handler type for interrupts that it
dispatches and the isa interrupt handler type for the interrupts that
it handles.
Reduce default value of pcicb_membase to 0x2000000 (from 0x4000000)
since this seems to be the lower bound used by many systems.
Submitted by: Mihoko Tanaka <m_tanaka@pa.yokogawa.co.jp>
successfully run linux netscape 2.0b3 with a QMAGIC ld.so and libc/libm
that I found on some linux machine that I _think_ is running slackware 3.0.
There are still problems.. ld.so claims the libraries are the wrong
format, but it still runs anyway.. :-/ The QMAGIC ld.so also screams
about needing ld.so.cache, and running a linux ldconfig is quite
educational. You soon learn to run "chroot /compat/linux /bin/ldconfig"
where ldconfig is living in /compat/linux/bin. :-]
(Lets just say that it puts loads of symlinks in /usr/lib otherwise :-)
(TTMAXHIWAT + OBUFSIZ + 100) in case someone changes OBUFSIZ. 200
was to allow 100 above high water for ordinary writes and another
100 for kernel printfs.
Increased the reserved output queue count from 512 to the maximum
output queue count. This prevents exhaustion of clists and increases
the output throughput for 8 cy lines by almost a factor of 2 (on
a system where there aren't many other open ttys so clists become
exhausted after about 4 active lines (or earlier if TTMAXHIWAT is
increased :-]).
ttwrite() behaves very badly when clists are exhausted:
(1) it sleeps on lbolt instead of on TSA_OLOWAT(tp).
This could be fixed adequately by sleeping on TSA_OLOWAT(tp).
The nonzero reserved count guaratees that space will become
available independent of other ttys, and a reserved count
of 512 is barely enough for efficiency.
(2) it drops output if space runs out in the middle of special
output processing. This is too hard to fix without hardening
the reserved count. The watermark processing guarantees that
space doesn't run out only if the advertised space is guaranteed.
Increasing the reserved output queue count defeats the point of
dynamic allocation of clists. Previously, about 2K of memory per
tty was reserved (the raw queue was already reserved). Now, about
3.5K is reserved. Reserving everything would take a whole 0.5K
more.
ttyflush() might be called from an interrupt handler. This fixes
panics in IXOFF mode at the cost of more failures to send the START
character to exit from IXOFF mode.
getmajorbyname() which were a better (sigh) temporary interface to
the going-away devswitches.
Note that SYSINIT()s to initialize the devswitches would be fatal
in syscons.c and pcvt_drv.c (and are bogus elsewhere) because they
get called independently of whether the device is attached; thus
devices that share a major clobber each other's devswitch entries
until the last one wins.
conf.c:
Removed stale #includes and comments.
LINT: add a couple of new/missing/undocumented options
files.i386: add linux code so that you can compile a kernel with static
linux emulation ("options LINUX")
i386/*: use #if defined(COMPAT_LINUX) || defined(LINUX) to enable static
support of linux emulation (just like "IBCS2" makes ibcs2 static)
The main thing this is going to make obvious, is that the LINUX code
(when compiled from LINT) has a lot of warnings, some of which dont look
too pleasant..
is possible to boot a kernel with an empty in-core MFS image, and have
it load the image from floppy directly. This is admittedly a hack and
would be better replaced by a self-loading ram-disk.
all the other bt_XXX() functions in i386/scsi/bt*.
This the important effect of forcing a link error if the user is
still using the old "vector btintr" which is dangerously wrong
after Justin's updates to the driver.
The correct isa vector line for the bt driver is "vector bt_isa_intr".
Justin mentioned this in the commit message and updated LINT and
GENERIC. This change is to enforce that.. :-)
seems to work hre just fine though I can't check every file
that changed due to limmited h/w, however I've checked enught to be petty
happy withe hte code..
WARNING... struct lkm[mumble] has changed
so it might be an idea to recompile any lkm related programs
up. The effect of this was that msync with a size would generally sync
1 page less than it should. This problem was brought to my attention
by Darrel Herbst <dherbst@gradin.cis.upenn.edu> and Ron Minnich
<rminnich@sarnoff.com>.
metadata and VBLK type devices. The code is currently mostly disabled,
and a work-around has been added to disabled attempted clustered writes
for VBLK type device buffers. Clustered write of meta-data is currently
a work in progress.
all buses.
Known problems:
-The PCI probe code has not been tested. Someone with a PCI Bt card will
have to validate it, but even if it is broken all cards the earlier version
of this driver found in ISA compatibility mode should still be found.
-Still missing the BT956 PCI ID, so it will be found as an ISA card until
someone suplies it.
-PCI interrupts go through an interrupt stub that returns an int until
we remove the edge-triggered PCI compatibiliity cruft.
-ISA interrupts go through an interrupt stub until they pass in (void *).
-The driver could support more mboxes and concurrent commands by allocating
structures separately and hanging them off the bt_data struct to get around
the 4K page limit. Someone with documentation should do this and also
enable tagged queuing.
can be found in ISA compatibility mode by the ISA driver, but since the
EISA and PCI probes are non-invasive, we prefer them to find the card first.
Since both EISA and PCI probes can rely on interrupts, enable them before
probing of any type is performed. All ISA probes are still "protected" by
splhigh().
allow one EISA/ISA/PCI/VL Buslogic controller to be probed. The driver
is almost fully dynamic. It just needs some kdc work and for the SCSI code
to stop passing unit numbers up in the scsi_xfer struct.
"getblk" hang. The B_WANTED flag was being cleared gratuitously,
also the optimization of gbincore for ignoring the B_INVAL flag was
incorrect. There is no place in the code where buffers are on the
hash list that are B_INVAL and not B_BUSY.
device must be configured. It's hard to tell whether a reset function
should be noreset or nullreset since reset functions are never called.
Most drivers use nullreset but noreset has the advantage of complaining
if somehow gets called).
Removed old aliases d_rdwr_t and d_ttycv_t for d_read_t/d_write_t and
d_devtotty_t.
Sorted declarations of switch functions into switch order.
Removed duplicated comments and declarations of nonexistent switch
functions.
It will need to be changed
but it's the better starting point..
also add '?' to wildcarding in SCSI identification of devices..
so we can catch all PIONEER CD 6??* devices instead of having
separate entries for the 600, 602, 604X, 624X etc..
it's getting so we should have a small regexp routine in the kernel
maybe just a little one.. matching CDX-6[0-9][0-9][ A-Z] would be better
there will be drastic changes in this
but this is the best starting point..
<sys/types.h> (if KERNEL is defined). This allows removing bogus
dependencies on vm stuff in several places (e.g., ddb) and stops
<vm_param.h> from depending on <vm_param.h>
Added declaration of boolean_t to <vm/vm.h> (if KERNEL is not
defined). It never belonged in <vm/vm_param.h>. Unfortunately,
it is required for some vm headers that are included by applications.
Deleted declarations of TRUE and FALSE from <vm/vm_param.h>. They
are defined in <sys/param.h> if KERNEL is defined and we'll soon
find out if any applications depend on them being defined in a vm
header.
Add five sysctl variables that you should probably never tweak.
net.arp.t_prune: 300
net.arp.t_keep: 1200
net.arp.t_down: 20
net.arp.maxtries: 5
net.arp.useloopback: 1
net.arp.proxyall: 0
(It's net.arp because arp isn't limited to inet, though our present
implementation surely is).
#include <sys/user> used to be self contained, but now it needs either
half a dozen VM specific includes beforehand (yuck, so much for
portability), or some horrible hack like this for user-mode only
applications.. The kind of stuff that needs this is the libkvm stuff,
w, ps, etc... I would welcome a better fix for this BTW.. :-)
(note: this is #ifndef KERNEL, so it shouldn't be re-polluting the kernel
space after it's been so painfully cleaned up...)
This is now in line with NetBSD as well..
Note that once this series of commits is finished, you must recompile
libkvm, then ps and maybe 'w'. If you are running the recently imported
sendmail-8.7, you should recompile that too (src/conf.c at least).
most devsw referenced functions are now static, as they are
in the same file as their devsw structure. I've also added DEVFS
support for nearly every device in the system, however
many of the devices have 'incorrect' names under DEVFS
because I couldn't quickly work out the correct naming conventions.
(but devfs won't be coming on line for a month or so anyhow so that doesn't
matter)
If you "OWN" a device which would normally have an entry in /dev
then search for the devfs_add_devsw() entries and munge to make them right..
check out similar devices to see what I might have done in them in you
can't see what's going on..
for a laugh compare conf.c conf.h defore and after... :)
I have not doen DEVFS entries for any DISKSLICE devices yet as that will be
a much more complicated job.. (pass 5 :)
pass 4 will be to make the devsw tables of type (cdevsw * )
rather than (cdevsw)
seems to work here..
complaints to the usual places.. :)
for the particular card in use. At the moment, I've set it to any of
the bt445S VLB cards (not the bt445C which apparently work) and the
bt5xx series (isa cards). The 742 and PCI cards should not need it. :-)
It may be useful to have something like this:
#ifndef BOUNCE_BUFFERS
if (bounce_buffers_required && more_than_16MB_ram)
panic("this card requires bounce buffers for more than 16MB ram!")
#endif
to get the definitions of TRUE and FALSE which happen to be defined in
a deeply nested include.
Added nearby #includes of <sys/conf.h> where appropriate.
single typedef) is now declared in <sys/types.h>.
This is the first of 4 commits that will remove some excessive
includes of vm*.h and user.h. The total speed improvement isn't
as large as I first thought. `make depend; make' for LINT only
improved from 2180 seconds to 2108 seconds user time.
is not real helpful since swapgeneric.c doesn't seem to be used, except
perhaps on a GENERIC kernel. (Sorry Paul.. :-)
I've moved it from swapgeneric.c to autoconf.c, since autoconf.c also deals
with dumpdev things. There may be a better place.....
as in <vm/vm_param.h> so that <vm/vm.h> doesn't have to be included
in kernel sources just to get the definitions of these fundamental
vm (;-) quantities.
but probably harmless. It's hard to tell because apparently no one runs
ity.
Fixed ity's d_reset entry. `nx' entries should never be used for existing
devices.
conf.c:
Moved a prototype to a better place.
Removed a stale #define.
sscselect(). Use the standard dummies nostrategy(), noread(),
nowrite() and noselect() instead.
sscread() and sscwrite() returned bogus errnos. It isn't possible
to return an error from a select routine so noselect() is just as
bogus as sscselect() (it's equivalent to nullselect()).
Fixed two cases of "=" that should have been "==" in card type comparison.
Simplified expression that checks for interface up/down.
Moved ed_ring_copy to before its first use so that it's inlined as intended.
Change mbuf allocation policy so that a received packet is stored in just
an mbuf header (no cluster) if it will fit in one.
Removed ifnet.if_init and ifnet.if_reset as they are generally unused.
Change the parameter passed to if_watchdog to be a ifnet * rather than
a unit number. All of this is an attempt to move toward not needing an
array of softc pointers (which is usually static in size) to point to
the driver softc.
if_ed.c:
Changed some of the argument passing to some functions to make a little
more sense.
if_ep.c, if_vx.c:
Killed completely bogus use of if_timer. It was being set in such a way
that the interface was being reset once per second (blech!).
Move a lot of variables home to their own code (In good time before xmas :-)
Introduce the string descrition of format.
Add a couple more functions to poke into these marvels, while I try to
decide what the correct interface should look like.
Next is adding vars on the fly, and sysctl looking at them too.
Removed a tine bit of defunct and #ifdefed notused code in swapgeneric.
Declared dsioctl() as static consistently. Note that both if_disc.c
and subr_diskslice.c use the same prefix `ds' and there is a name
conflict for dsioctl().
cd9660_rrip.c:
Added lots of bogus casts to hide type errors exposed by the prototypes.
(Different structs are assumed to have a common prefix.)
cd9660_vnops.c:
Finished staticizing.
Protected them with `#ifdef KERNEL' so that <sys/queue.h> is valid C++.
Added the necessary #includes of <sys/queue.h>.
These functions are bogus and should be replaced by the queue macros.
Staticized some functions.
__purified some functions. Some functions were bogusly declared as
returning `const'. This hasn't done anything since gcc-2.5. For
later versions of gcc, the equivalent is __attribute__((const)) at
the end of function declarations.
Obtained from: NetBSD as well (He submitted it there too)
make sure that teh shm region is beyond the sum of the text and data segs
as it was big progs could collide with the shm region.
but also IT ACTUALLY WORKS!
FreeBSD with options JREMOD now runs with no entries in the devsw tables
prior to the devices puting their own entries there..
Thanks to bde and terry for thoughts and comments.
next stop 'Real' devfs support in devices.
- Don't print out meaningless iCOMP numbers, those are for droids.
- Use a shorter wait to determine clock rate to avoid deficiencies
in DELAY().
- Use a fixed-point representation with 8 bits of fraction to store
the rate and rationalize the variable name. It would be
possible to use even more fraction if it turns out to be
worthwhile (I rather doubt it).
The question of source code arrangement remains unaddressed.
conflicted with S3 graphics cards. Now users should put sio[2-3]
in the config file if the hardware exisst, even if the probe is
certain to fail due to an interrupt conflict. Otherwise, ports
sharing the interrupt may fail the probe if the system is warm
booted while sio[2-3] are active (perhaps under another OS). The
same problem for nonstandard ports is now handled better than
before.
add a few safety checks in specfs because
now it's possible to get entries in [cd]devsw[] which are ALL NULL
so it's better to discover this BEFORE jumping into the d_open() entry..
more check to come later.. this getsthe code to the stage where I
can start testing it, even if I haven't caught every little error case...
I guess I'll find them quick enough..
That's EVERY SINGLE driver that has an entry in conf.c..
my next trick will be to define cdevsw[] and bdevsw[]
as empty arrays and remove all those DAMNED defines as well..
Each of these drivers has a SYSINIT linker set entry
that comes in very early.. and asks teh driver to add it's own
entry to the two devsw[] tables.
some slight reworking of the commits from yesterday (added the SYSINIT
stuff and some usually wrong but token DEVFS entries to all these
devices.
BTW does anyone know where the 'ata' entries in conf.c actually reside?
seems we don't actually have a 'ataopen() etc...
If you want to add a new device in conf.c
please make sure I know
so I can keep it up to date too..
as before, this is all dependent on #if defined(JREMOD)
(and #ifdef DEVFS in parts)
o Add signed/unsigned functionality to the matrox meteor device driver.
o Apply a few fixes to the sound driver.
o Add a ``SPIGOT_UNSECURE'' compile time definition so, if one defines
SPIGOT_UNSECURE in their conf file, then they can use the spigot w/o
root. There is a warning that this allows users access to the IO
page which is probably not secure.
Submitted by: james
totally dynamic. (the first was about 7 weeeks ago)
this is only the devices in i386/isa
I'll do more tomorrow.
they're completely masked by #ifdef JREMOD at this stage...
the eventual aim is that every driver will do a SYSINIT
at startup BEFORE the probes, which will effectively
link it into the devsw tables etc.
If I'd thought about it more I'd have put that in in this set (damn)
The ioconf lines generated by config will also end up in the
device's own scope as well, so ioconf.c will eventually be gutted
the SYSINIT call to the driver will include a phase where the
driver links it's ioconf line into a chain of such. when this phase is done
then the user can modify them with the boot: -c
config menu if he wants, just like now..
config will put the config lines out in the .h file
(e.g. in aha.h will be the addresses for the aha driver to look.)
as I said this is a very small first step..
the aim of THIS set of edits is to not have to edit conf.c at all when
adding a new device.. the tabe will be a simple skeleton..
when this is done, it will allow other changes to be made,
all teh time still having a fully working kernel tree,
but the logical outcome is the complete REMOVAL of the devsw tables.
By the end of this, linked in drivers will be exactly the same as
run-time loaded drivers, except they JUST HAPPEN to already be linked
and present at startup..
the SYSINIT calls will be the equivalent of the "init" call
made to a newly loaded driver in every respect.
For this edit,
each of the files has the following code inserted into it:
obviously, tailored to suit..
----------------------somewhere at the top:
#ifdef JREMOD
#include <sys/conf.h>
#define CDEV_MAJOR 13
#define BDEV_MAJOR 4
static void sd_devsw_install();
#endif /*JREMOD */
---------------------somewhere that's run during bootup: EVENTUALLY a SYSINIT
#ifdef JREMOD
sd_devsw_install();
#endif /*JREMOD*/
-----------------------at the bottom:
#ifdef JREMOD
struct bdevsw sd_bdevsw =
{ sdopen, sdclose, sdstrategy, sdioctl, /*4*/
sddump, sdsize, 0 };
struct cdevsw sd_cdevsw =
{ sdopen, sdclose, rawread, rawwrite, /*13*/
sdioctl, nostop, nullreset, nodevtotty,/* sd */
seltrue, nommap, sdstrategy };
static sd_devsw_installed = 0;
static void sd_devsw_install()
{
dev_t descript;
if( ! sd_devsw_installed ) {
descript = makedev(CDEV_MAJOR,0);
cdevsw_add(&descript,&sd_cdevsw,NULL);
#if defined(BDEV_MAJOR)
descript = makedev(BDEV_MAJOR,0);
bdevsw_add(&descript,&sd_bdevsw,NULL);
#endif /*BDEV_MAJOR*/
sd_devsw_installed = 1;
}
}
#endif /* JREMOD */
totally dynamic.
this is only the devices in i386/isa
I'll do more tomorrow.
they're completely masked by #ifdef JREMOD at this stage...
the eventual aim is that every driver will do a SYSINIT
at startup BEFORE the probes, which will effectively
link it into the devsw tables etc.
If I'd thought about it more I'd have put that in in this set (damn)
The ioconf lines generated by config will also end up in the
device's own scope as well, so ioconf.c will eventually be gutted
the SYSINIT call to the driver will include a phase where the
driver links it's ioconf line into a chain of such. when this phase is done
then the user can modify them with the boot: -c
config menu if he wants, just like now..
config will put the config lines out in the .h file
(e.g. in aha.h will be the addresses for the aha driver to look.)
as I said this is a very small first step..
the aim of THIS set of edits is to not have to edit conf.c at all when
adding a new device.. the tabe will be a simple skeleton..
when this is done, it will allow other changes to be made,
all teh time still having a fully working kernel tree,
but the logical outcome is the complete REMOVAL of the devsw tables.
By the end of this, linked in drivers will be exactly the same as
run-time loaded drivers, except they JUST HAPPEN to already be linked
and present at startup..
the SYSINIT calls will be the equivalent of the "init" call
made to a newly loaded driver in every respect.
For this edit,
each of the files has the following code inserted into it:
obviously, tailored to suit..
----------------------somewhere at the top:
#ifdef JREMOD
#include <sys/conf.h>
#define CDEV_MAJOR 13
#define BDEV_MAJOR 4
static void sd_devsw_install();
#endif /*JREMOD */
---------------------somewhere that's run during bootup: EVENTUALLY a SYSINIT
#ifdef JREMOD
sd_devsw_install();
#endif /*JREMOD*/
-----------------------at the bottom:
#ifdef JREMOD
struct bdevsw sd_bdevsw =
{ sdopen, sdclose, sdstrategy, sdioctl, /*4*/
sddump, sdsize, 0 };
struct cdevsw sd_cdevsw =
{ sdopen, sdclose, rawread, rawwrite, /*13*/
sdioctl, nostop, nullreset, nodevtotty,/* sd */
seltrue, nommap, sdstrategy };
static sd_devsw_installed = 0;
static void sd_devsw_install()
{
dev_t descript;
if( ! sd_devsw_installed ) {
descript = makedev(CDEV_MAJOR,0);
cdevsw_add(&descript,&sd_cdevsw,NULL);
#if defined(BDEV_MAJOR)
descript = makedev(BDEV_MAJOR,0);
bdevsw_add(&descript,&sd_bdevsw,NULL);
#endif /*BDEV_MAJOR*/
sd_devsw_installed = 1;
}
}
#endif /* JREMOD */
self-decompressing ram disk that I'm fiddling with..
(Note, this depends on the various syscalls having correctly set uio_segflag
before calling physio - I've checked and they look correct.)
What was happening, was that the main mfs loop was sleeping, and when it was
being awoken by a wakeup when it was supposed to process some IO requests.
The problem was that if it was being woken out of the tsleep() by a signal
at shutdown, it was going straight into dounmount() without servicing any
pending IO requests, causing dounmount() to fail because there were busy
buffers (and they could not be "processed" because the processing loop was
trying to unmount rather than dispatching into mfs_doio()).
This (dare I say it :-) appears to be a layering problem....
