Remove Softdep.

Softdep has been a no-op for some time now, this removes it to get
it out of the way.

Flensing mostly done in Talinn, with some help from krw@

ok deraadt@

Fix comments longer than 80 column.

ok miod@

Relax the grip the dead hand of QBUS (a.k.a. MAXPHYS) has on the
throat of physio().

Allows individual devices to eventually be modified to support larger
physio() (a.k.a. 'raw') i/o's if they prove capable of them.

No immediate functional change.

Tested for many weeks by and ok robert@.

Use long for temporary variables handling b_bcount values in physio().

Add a KASSERTMSG() to check that strategy functions didn't screw
b_resid up too much.

ok beck@ tedu@

physio() just needs to check for b_bcount overflow. Let the provided
minphys() function check for MAXPHYS.

Feedback from tedu@ kettenis@ dlg@

ok cheloha@, robert@, jmatthew@ as part of larger diff

Convert infinite sleeps to {m,t}sleep_nsec(9).

ok visa@

Remove some includes include-what-you-use claims don't
have any direct symbols used. Tested for indirect use by compiling
amd64/i386/sparc64 kernels.

ok tedu@ deraadt@

Replace a plethora of historical protection options with just
PROT_NONE, PROT_READ, PROT_WRITE, and PROT_EXEC from mman.h.
PROT_MASK is introduced as the one true way of extracting those bits.
Remove UVM_ADV_* wrapper, using the standard names.
ok doug guenther kettenis

remove uneeded proc.h includes
ok mpi@ kspillner@

KASSERTMSG(9): New kernel assertion with message

KASSERT() is annoying as it only prints the expression as a string. If you
(developers) want to know a little more information, you have to do:

#ifdef DIAGNOSTIC
if (bad)
panic(...);
#endif

KASSERTMSG() replaces it into a single line:

KASSERTMSG(!bad, ...);

Taken from NetBSD.

(There is a concern that KASSERT() messages are too long; consume more memory,
and not friendly for small monitors. This have to be considered & revisited
later.)

"Like" from henning@
Man page review & advices from jmc@ and schwarze@

Revert vnd(4) to its older less strict behavior.

Verified by deraadt@ to fix distrib/sgi/iso.

set flags correctly in physio - now with actual clearing of B_DONE and B_ERROR
between re-use of the same buffer to re-call strategy() - large dd's now
work again which were broken by my last commit of this.
ok krw@

back out last change; it caused dd on usb drive to lock the machine.

ok oga@

stylistic clean of buffer flags assignment to be more obviously not a
destructive assignment of flags.
ok oga@

If uvm_vslock_device() fails, just exit the loop. None of the
after_unlock code is needed if this happens, and running it was even
wrong because we weren't setting b_resid, so uio_offset (and higher
up, f_offset) would be incorrectly adjusted.

Discussed with deraadt@.

Enforce that raw disk I/O is only initiated on block boundaries in
physio().

ok deraadt@

because bufs now come out of a pool rather than a global list of staticly
allocated structs, they wont be shared. we dont have to do the silly
B_WANTED dance anymore. get rid of a bunch of comments that talk about
how we're stealing from the swapper, which is a lie now.

tweaks from oga@ scrutiny by matthew@
ok miod@ oga@ thib@ beck@

All users of physio(9) now pass NULL as the buf pointer argument, so
no point in keeping it around.

"i like this" thib@ (a while back); ok krw@ and oga@; reminder to
update the man page and tweaks jmc@

Implement vs{,un}lock_device and use it for physio.

Just like normal vs{,un}lock, but in case the pages we get are not dma
accessible, we bounce them, if they are dma acessible, the functions behave
exactly like normal vslock. The plan for the future is to have fault_wire
allocate dma acessible pages so that we don't need to bounce (especially
in cases where the same buffer is reused for physio over and over again),
but for now, keep it as simple as possible.

No need to do the splbio/splx dance around the pool_get in
getphysbuf() as the bufpool is already protected.

From Matthew Dempsky <matthew at dempsky dot org>
OK blambert@

obvious shrinkage.

ok krw@, miod@.

Dynamic buffer cache. Initial diff from mickey@, okay art@ beck@ toby@
deraadt@ dlg@.

'tranfer' -> 'transfer' in comments.

Kernel stack can be swapped. This means that stuff that's on the stack
should never be referenced outside the context of the process to which
this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
doesn't follow the rules here. Instead of trying to track down all
offenders and fix this hairy situation, it makes much more sense
to not swap kernel stacks.

From art@, tested by many some time ago.

typos; from bret lambert

Don't panic for pathological i/o sizes unless minphys() really is
broken. Eliminate an unneeded variable and potential conversion issues
in SCIOCCOMMAND code before calling physio.

Similar to what NetBSD does.

Fixes cdda2wav vs "Billie Holiday - Songs for Distingue Lovers"
problem noted by Alexandre Ratchov. Tested by Alexandre.

ok marco@ pedro@ deraadt@ mickey@

ansi/deregister.
'go for it' deraadt@

typos from Jonathon Gray;

remove caddr_t casts. it's just silly to cast something when the function
takes a void *. convert uiomove to take a void * as well. ok deraadt@

Remove the advertising clause in the UCB license which Berkeley
rescinded 22 July 1999. Proofed by myself and Theo.

First round of __P removal in sys

branches: 1.18.2;
No need for uvm_useracc here now. vslock catches those problems.

Sync in more uvm changes from NetBSD.
This time we're getting rid of KERN_* and VM_PAGER_* error codes and
use errnos instead.

Remove creds from struct buf, move the creds that nfs need into the nfs node.
While in the area, convert nfs node allocation from malloc to pool and do
some cleanups.
Based on the UBC changes in NetBSD. niklas@ ok.

Use bufpool, not malloc.

Replace inclusion of <vm/foo.h> with the correct <uvm/bar.h> when necessary.
(Look ma, I might have broken the tree)

remove old vm

Appease gcc by not using void pointers in arithmetic operations; art@ ok

KNF

Check for failure now that uvm_vslock can return one. Before this, there was
a risk that we could crash when doing physio/sysctl when the system was
completly out of RAM and swap.

Use PHOLD/PRELE instead of manually fiddling with p_holdcnt.

Bring in some new UVM code from NetBSD (not current).

- Introduce a new type of map that are interrupt safe and never allow faults
in them. mb_map and kmem_map are made intrsafe.
- Add "access protection" to uvm_vslock (to be passed down to uvm_fault and
later to pmap_enter).
- madvise(2) now works.
- various cleanups.

branches: 1.7.2;
Init the buffer dependency list.

s/memset/bzero/; art@ ok

mostly name changes and allocation chnges for uvm

indent properly

from netbsd: check user access rights only if UIO_USERSPACE

From NetBSD: 960217 merge

branches: 1.1.1;
Initial revision

Fix comments longer than 80 column.

ok miod@

Relax the grip the dead hand of QBUS (a.k.a. MAXPHYS) has on the
throat of physio().

Allows individual devices to eventually be modified to support larger
physio() (a.k.a. 'raw') i/o's if they prove capable of them.

No immediate functional change.

Tested for many weeks by and ok robert@.

Use long for temporary variables handling b_bcount values in physio().

Add a KASSERTMSG() to check that strategy functions didn't screw
b_resid up too much.

ok beck@ tedu@

physio() just needs to check for b_bcount overflow. Let the provided
minphys() function check for MAXPHYS.

Feedback from tedu@ kettenis@ dlg@

ok cheloha@, robert@, jmatthew@ as part of larger diff

Convert infinite sleeps to {m,t}sleep_nsec(9).

ok visa@

Remove some includes include-what-you-use claims don't
have any direct symbols used. Tested for indirect use by compiling
amd64/i386/sparc64 kernels.

ok tedu@ deraadt@

Replace a plethora of historical protection options with just
PROT_NONE, PROT_READ, PROT_WRITE, and PROT_EXEC from mman.h.
PROT_MASK is introduced as the one true way of extracting those bits.
Remove UVM_ADV_* wrapper, using the standard names.
ok doug guenther kettenis

remove uneeded proc.h includes
ok mpi@ kspillner@

KASSERTMSG(9): New kernel assertion with message

KASSERT() is annoying as it only prints the expression as a string. If you
(developers) want to know a little more information, you have to do:

#ifdef DIAGNOSTIC
if (bad)
panic(...);
#endif

KASSERTMSG() replaces it into a single line:

KASSERTMSG(!bad, ...);

Taken from NetBSD.

(There is a concern that KASSERT() messages are too long; consume more memory,
and not friendly for small monitors. This have to be considered & revisited
later.)

"Like" from henning@
Man page review & advices from jmc@ and schwarze@

Revert vnd(4) to its older less strict behavior.

Verified by deraadt@ to fix distrib/sgi/iso.

set flags correctly in physio - now with actual clearing of B_DONE and B_ERROR
between re-use of the same buffer to re-call strategy() - large dd's now
work again which were broken by my last commit of this.
ok krw@

back out last change; it caused dd on usb drive to lock the machine.

ok oga@

stylistic clean of buffer flags assignment to be more obviously not a
destructive assignment of flags.
ok oga@

If uvm_vslock_device() fails, just exit the loop. None of the
after_unlock code is needed if this happens, and running it was even
wrong because we weren't setting b_resid, so uio_offset (and higher
up, f_offset) would be incorrectly adjusted.

Discussed with deraadt@.

Enforce that raw disk I/O is only initiated on block boundaries in
physio().

ok deraadt@

because bufs now come out of a pool rather than a global list of staticly
allocated structs, they wont be shared. we dont have to do the silly
B_WANTED dance anymore. get rid of a bunch of comments that talk about
how we're stealing from the swapper, which is a lie now.

tweaks from oga@ scrutiny by matthew@
ok miod@ oga@ thib@ beck@

All users of physio(9) now pass NULL as the buf pointer argument, so
no point in keeping it around.

"i like this" thib@ (a while back); ok krw@ and oga@; reminder to
update the man page and tweaks jmc@

Implement vs{,un}lock_device and use it for physio.

Just like normal vs{,un}lock, but in case the pages we get are not dma
accessible, we bounce them, if they are dma acessible, the functions behave
exactly like normal vslock. The plan for the future is to have fault_wire
allocate dma acessible pages so that we don't need to bounce (especially
in cases where the same buffer is reused for physio over and over again),
but for now, keep it as simple as possible.

No need to do the splbio/splx dance around the pool_get in
getphysbuf() as the bufpool is already protected.

From Matthew Dempsky <matthew at dempsky dot org>
OK blambert@

obvious shrinkage.

ok krw@, miod@.

Dynamic buffer cache. Initial diff from mickey@, okay art@ beck@ toby@
deraadt@ dlg@.

'tranfer' -> 'transfer' in comments.

Kernel stack can be swapped. This means that stuff that's on the stack
should never be referenced outside the context of the process to which
this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
doesn't follow the rules here. Instead of trying to track down all
offenders and fix this hairy situation, it makes much more sense
to not swap kernel stacks.

From art@, tested by many some time ago.

typos; from bret lambert

Don't panic for pathological i/o sizes unless minphys() really is
broken. Eliminate an unneeded variable and potential conversion issues
in SCIOCCOMMAND code before calling physio.

Similar to what NetBSD does.

Fixes cdda2wav vs "Billie Holiday - Songs for Distingue Lovers"
problem noted by Alexandre Ratchov. Tested by Alexandre.

ok marco@ pedro@ deraadt@ mickey@

ansi/deregister.
'go for it' deraadt@

typos from Jonathon Gray;

remove caddr_t casts. it's just silly to cast something when the function
takes a void *. convert uiomove to take a void * as well. ok deraadt@

Remove the advertising clause in the UCB license which Berkeley
rescinded 22 July 1999. Proofed by myself and Theo.

First round of __P removal in sys

branches: 1.18.2;
No need for uvm_useracc here now. vslock catches those problems.

Sync in more uvm changes from NetBSD.
This time we're getting rid of KERN_* and VM_PAGER_* error codes and
use errnos instead.

Remove creds from struct buf, move the creds that nfs need into the nfs node.
While in the area, convert nfs node allocation from malloc to pool and do
some cleanups.
Based on the UBC changes in NetBSD. niklas@ ok.

Use bufpool, not malloc.

Replace inclusion of <vm/foo.h> with the correct <uvm/bar.h> when necessary.
(Look ma, I might have broken the tree)

remove old vm

Appease gcc by not using void pointers in arithmetic operations; art@ ok

KNF

Check for failure now that uvm_vslock can return one. Before this, there was
a risk that we could crash when doing physio/sysctl when the system was
completly out of RAM and swap.

Use PHOLD/PRELE instead of manually fiddling with p_holdcnt.

Bring in some new UVM code from NetBSD (not current).

- Introduce a new type of map that are interrupt safe and never allow faults
in them. mb_map and kmem_map are made intrsafe.
- Add "access protection" to uvm_vslock (to be passed down to uvm_fault and
later to pmap_enter).
- madvise(2) now works.
- various cleanups.

branches: 1.7.2;
Init the buffer dependency list.

s/memset/bzero/; art@ ok

mostly name changes and allocation chnges for uvm

indent properly

from netbsd: check user access rights only if UIO_USERSPACE

From NetBSD: 960217 merge

branches: 1.1.1;
Initial revision

Relax the grip the dead hand of QBUS (a.k.a. MAXPHYS) has on the
throat of physio().

Allows individual devices to eventually be modified to support larger
physio() (a.k.a. 'raw') i/o's if they prove capable of them.

No immediate functional change.

Tested for many weeks by and ok robert@.

Use long for temporary variables handling b_bcount values in physio().

Add a KASSERTMSG() to check that strategy functions didn't screw
b_resid up too much.

ok beck@ tedu@

physio() just needs to check for b_bcount overflow. Let the provided
minphys() function check for MAXPHYS.

Feedback from tedu@ kettenis@ dlg@

ok cheloha@, robert@, jmatthew@ as part of larger diff

Convert infinite sleeps to {m,t}sleep_nsec(9).

ok visa@

Remove some includes include-what-you-use claims don't
have any direct symbols used. Tested for indirect use by compiling
amd64/i386/sparc64 kernels.

ok tedu@ deraadt@

Replace a plethora of historical protection options with just
PROT_NONE, PROT_READ, PROT_WRITE, and PROT_EXEC from mman.h.
PROT_MASK is introduced as the one true way of extracting those bits.
Remove UVM_ADV_* wrapper, using the standard names.
ok doug guenther kettenis

remove uneeded proc.h includes
ok mpi@ kspillner@

KASSERTMSG(9): New kernel assertion with message

KASSERT() is annoying as it only prints the expression as a string. If you
(developers) want to know a little more information, you have to do:

#ifdef DIAGNOSTIC
if (bad)
panic(...);
#endif

KASSERTMSG() replaces it into a single line:

KASSERTMSG(!bad, ...);

Taken from NetBSD.

(There is a concern that KASSERT() messages are too long; consume more memory,
and not friendly for small monitors. This have to be considered & revisited
later.)

"Like" from henning@
Man page review & advices from jmc@ and schwarze@

Revert vnd(4) to its older less strict behavior.

Verified by deraadt@ to fix distrib/sgi/iso.

set flags correctly in physio - now with actual clearing of B_DONE and B_ERROR
between re-use of the same buffer to re-call strategy() - large dd's now
work again which were broken by my last commit of this.
ok krw@

back out last change; it caused dd on usb drive to lock the machine.

ok oga@

stylistic clean of buffer flags assignment to be more obviously not a
destructive assignment of flags.
ok oga@

If uvm_vslock_device() fails, just exit the loop. None of the
after_unlock code is needed if this happens, and running it was even
wrong because we weren't setting b_resid, so uio_offset (and higher
up, f_offset) would be incorrectly adjusted.

Discussed with deraadt@.

Enforce that raw disk I/O is only initiated on block boundaries in
physio().

ok deraadt@

because bufs now come out of a pool rather than a global list of staticly
allocated structs, they wont be shared. we dont have to do the silly
B_WANTED dance anymore. get rid of a bunch of comments that talk about
how we're stealing from the swapper, which is a lie now.

tweaks from oga@ scrutiny by matthew@
ok miod@ oga@ thib@ beck@

All users of physio(9) now pass NULL as the buf pointer argument, so
no point in keeping it around.

"i like this" thib@ (a while back); ok krw@ and oga@; reminder to
update the man page and tweaks jmc@

Implement vs{,un}lock_device and use it for physio.

Just like normal vs{,un}lock, but in case the pages we get are not dma
accessible, we bounce them, if they are dma acessible, the functions behave
exactly like normal vslock. The plan for the future is to have fault_wire
allocate dma acessible pages so that we don't need to bounce (especially
in cases where the same buffer is reused for physio over and over again),
but for now, keep it as simple as possible.

No need to do the splbio/splx dance around the pool_get in
getphysbuf() as the bufpool is already protected.

From Matthew Dempsky <matthew at dempsky dot org>
OK blambert@

obvious shrinkage.

ok krw@, miod@.

Dynamic buffer cache. Initial diff from mickey@, okay art@ beck@ toby@
deraadt@ dlg@.

'tranfer' -> 'transfer' in comments.

Kernel stack can be swapped. This means that stuff that's on the stack
should never be referenced outside the context of the process to which
this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
doesn't follow the rules here. Instead of trying to track down all
offenders and fix this hairy situation, it makes much more sense
to not swap kernel stacks.

From art@, tested by many some time ago.

typos; from bret lambert

Don't panic for pathological i/o sizes unless minphys() really is
broken. Eliminate an unneeded variable and potential conversion issues
in SCIOCCOMMAND code before calling physio.

Similar to what NetBSD does.

Fixes cdda2wav vs "Billie Holiday - Songs for Distingue Lovers"
problem noted by Alexandre Ratchov. Tested by Alexandre.

ok marco@ pedro@ deraadt@ mickey@

ansi/deregister.
'go for it' deraadt@

typos from Jonathon Gray;

remove caddr_t casts. it's just silly to cast something when the function
takes a void *. convert uiomove to take a void * as well. ok deraadt@

Remove the advertising clause in the UCB license which Berkeley
rescinded 22 July 1999. Proofed by myself and Theo.

First round of __P removal in sys

branches: 1.18.2;
No need for uvm_useracc here now. vslock catches those problems.

Sync in more uvm changes from NetBSD.
This time we're getting rid of KERN_* and VM_PAGER_* error codes and
use errnos instead.

Remove creds from struct buf, move the creds that nfs need into the nfs node.
While in the area, convert nfs node allocation from malloc to pool and do
some cleanups.
Based on the UBC changes in NetBSD. niklas@ ok.

Use bufpool, not malloc.

Replace inclusion of <vm/foo.h> with the correct <uvm/bar.h> when necessary.
(Look ma, I might have broken the tree)

remove old vm

Appease gcc by not using void pointers in arithmetic operations; art@ ok

KNF

Check for failure now that uvm_vslock can return one. Before this, there was
a risk that we could crash when doing physio/sysctl when the system was
completly out of RAM and swap.

Use PHOLD/PRELE instead of manually fiddling with p_holdcnt.

Bring in some new UVM code from NetBSD (not current).

- Introduce a new type of map that are interrupt safe and never allow faults
in them. mb_map and kmem_map are made intrsafe.
- Add "access protection" to uvm_vslock (to be passed down to uvm_fault and
later to pmap_enter).
- madvise(2) now works.
- various cleanups.

branches: 1.7.2;
Init the buffer dependency list.

s/memset/bzero/; art@ ok

mostly name changes and allocation chnges for uvm

indent properly

from netbsd: check user access rights only if UIO_USERSPACE

From NetBSD: 960217 merge

branches: 1.1.1;
Initial revision

Use long for temporary variables handling b_bcount values in physio().

Add a KASSERTMSG() to check that strategy functions didn't screw
b_resid up too much.

ok beck@ tedu@

physio() just needs to check for b_bcount overflow. Let the provided
minphys() function check for MAXPHYS.

Feedback from tedu@ kettenis@ dlg@

ok cheloha@, robert@, jmatthew@ as part of larger diff

Convert infinite sleeps to {m,t}sleep_nsec(9).

ok visa@

Remove some includes include-what-you-use claims don't
have any direct symbols used. Tested for indirect use by compiling
amd64/i386/sparc64 kernels.

ok tedu@ deraadt@

Replace a plethora of historical protection options with just
PROT_NONE, PROT_READ, PROT_WRITE, and PROT_EXEC from mman.h.
PROT_MASK is introduced as the one true way of extracting those bits.
Remove UVM_ADV_* wrapper, using the standard names.
ok doug guenther kettenis

remove uneeded proc.h includes
ok mpi@ kspillner@

KASSERTMSG(9): New kernel assertion with message

KASSERT() is annoying as it only prints the expression as a string. If you
(developers) want to know a little more information, you have to do:

#ifdef DIAGNOSTIC
if (bad)
panic(...);
#endif

KASSERTMSG() replaces it into a single line:

KASSERTMSG(!bad, ...);

Taken from NetBSD.

(There is a concern that KASSERT() messages are too long; consume more memory,
and not friendly for small monitors. This have to be considered & revisited
later.)

"Like" from henning@
Man page review & advices from jmc@ and schwarze@

Revert vnd(4) to its older less strict behavior.

Verified by deraadt@ to fix distrib/sgi/iso.

set flags correctly in physio - now with actual clearing of B_DONE and B_ERROR
between re-use of the same buffer to re-call strategy() - large dd's now
work again which were broken by my last commit of this.
ok krw@

back out last change; it caused dd on usb drive to lock the machine.

ok oga@

stylistic clean of buffer flags assignment to be more obviously not a
destructive assignment of flags.
ok oga@

If uvm_vslock_device() fails, just exit the loop. None of the
after_unlock code is needed if this happens, and running it was even
wrong because we weren't setting b_resid, so uio_offset (and higher
up, f_offset) would be incorrectly adjusted.

Discussed with deraadt@.

Enforce that raw disk I/O is only initiated on block boundaries in
physio().

ok deraadt@

because bufs now come out of a pool rather than a global list of staticly
allocated structs, they wont be shared. we dont have to do the silly
B_WANTED dance anymore. get rid of a bunch of comments that talk about
how we're stealing from the swapper, which is a lie now.

tweaks from oga@ scrutiny by matthew@
ok miod@ oga@ thib@ beck@

All users of physio(9) now pass NULL as the buf pointer argument, so
no point in keeping it around.

"i like this" thib@ (a while back); ok krw@ and oga@; reminder to
update the man page and tweaks jmc@

Implement vs{,un}lock_device and use it for physio.

Just like normal vs{,un}lock, but in case the pages we get are not dma
accessible, we bounce them, if they are dma acessible, the functions behave
exactly like normal vslock. The plan for the future is to have fault_wire
allocate dma acessible pages so that we don't need to bounce (especially
in cases where the same buffer is reused for physio over and over again),
but for now, keep it as simple as possible.

No need to do the splbio/splx dance around the pool_get in
getphysbuf() as the bufpool is already protected.

From Matthew Dempsky <matthew at dempsky dot org>
OK blambert@

obvious shrinkage.

ok krw@, miod@.

Dynamic buffer cache. Initial diff from mickey@, okay art@ beck@ toby@
deraadt@ dlg@.

'tranfer' -> 'transfer' in comments.

Kernel stack can be swapped. This means that stuff that's on the stack
should never be referenced outside the context of the process to which
this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
doesn't follow the rules here. Instead of trying to track down all
offenders and fix this hairy situation, it makes much more sense
to not swap kernel stacks.

From art@, tested by many some time ago.

typos; from bret lambert

Don't panic for pathological i/o sizes unless minphys() really is
broken. Eliminate an unneeded variable and potential conversion issues
in SCIOCCOMMAND code before calling physio.

Similar to what NetBSD does.

Fixes cdda2wav vs "Billie Holiday - Songs for Distingue Lovers"
problem noted by Alexandre Ratchov. Tested by Alexandre.

ok marco@ pedro@ deraadt@ mickey@

ansi/deregister.
'go for it' deraadt@

typos from Jonathon Gray;

remove caddr_t casts. it's just silly to cast something when the function
takes a void *. convert uiomove to take a void * as well. ok deraadt@

Remove the advertising clause in the UCB license which Berkeley
rescinded 22 July 1999. Proofed by myself and Theo.

First round of __P removal in sys

branches: 1.18.2;
No need for uvm_useracc here now. vslock catches those problems.

Sync in more uvm changes from NetBSD.
This time we're getting rid of KERN_* and VM_PAGER_* error codes and
use errnos instead.

Remove creds from struct buf, move the creds that nfs need into the nfs node.
While in the area, convert nfs node allocation from malloc to pool and do
some cleanups.
Based on the UBC changes in NetBSD. niklas@ ok.

Use bufpool, not malloc.

Replace inclusion of <vm/foo.h> with the correct <uvm/bar.h> when necessary.
(Look ma, I might have broken the tree)

remove old vm

Appease gcc by not using void pointers in arithmetic operations; art@ ok

KNF

Check for failure now that uvm_vslock can return one. Before this, there was
a risk that we could crash when doing physio/sysctl when the system was
completly out of RAM and swap.

Use PHOLD/PRELE instead of manually fiddling with p_holdcnt.

Bring in some new UVM code from NetBSD (not current).

- Introduce a new type of map that are interrupt safe and never allow faults
in them. mb_map and kmem_map are made intrsafe.
- Add "access protection" to uvm_vslock (to be passed down to uvm_fault and
later to pmap_enter).
- madvise(2) now works.
- various cleanups.

branches: 1.7.2;
Init the buffer dependency list.

s/memset/bzero/; art@ ok

mostly name changes and allocation chnges for uvm

indent properly

from netbsd: check user access rights only if UIO_USERSPACE

From NetBSD: 960217 merge

branches: 1.1.1;
Initial revision

physio() just needs to check for b_bcount overflow. Let the provided
minphys() function check for MAXPHYS.

Feedback from tedu@ kettenis@ dlg@

ok cheloha@, robert@, jmatthew@ as part of larger diff

Convert infinite sleeps to {m,t}sleep_nsec(9).

ok visa@

Remove some includes include-what-you-use claims don't
have any direct symbols used. Tested for indirect use by compiling
amd64/i386/sparc64 kernels.

ok tedu@ deraadt@

Replace a plethora of historical protection options with just
PROT_NONE, PROT_READ, PROT_WRITE, and PROT_EXEC from mman.h.
PROT_MASK is introduced as the one true way of extracting those bits.
Remove UVM_ADV_* wrapper, using the standard names.
ok doug guenther kettenis

remove uneeded proc.h includes
ok mpi@ kspillner@

KASSERTMSG(9): New kernel assertion with message

KASSERT() is annoying as it only prints the expression as a string. If you
(developers) want to know a little more information, you have to do:

#ifdef DIAGNOSTIC
if (bad)
panic(...);
#endif

KASSERTMSG() replaces it into a single line:

KASSERTMSG(!bad, ...);

Taken from NetBSD.

(There is a concern that KASSERT() messages are too long; consume more memory,
and not friendly for small monitors. This have to be considered & revisited
later.)

"Like" from henning@
Man page review & advices from jmc@ and schwarze@

Revert vnd(4) to its older less strict behavior.

Verified by deraadt@ to fix distrib/sgi/iso.

set flags correctly in physio - now with actual clearing of B_DONE and B_ERROR
between re-use of the same buffer to re-call strategy() - large dd's now
work again which were broken by my last commit of this.
ok krw@

back out last change; it caused dd on usb drive to lock the machine.

ok oga@

stylistic clean of buffer flags assignment to be more obviously not a
destructive assignment of flags.
ok oga@

If uvm_vslock_device() fails, just exit the loop. None of the
after_unlock code is needed if this happens, and running it was even
wrong because we weren't setting b_resid, so uio_offset (and higher
up, f_offset) would be incorrectly adjusted.

Discussed with deraadt@.

Enforce that raw disk I/O is only initiated on block boundaries in
physio().

ok deraadt@

because bufs now come out of a pool rather than a global list of staticly
allocated structs, they wont be shared. we dont have to do the silly
B_WANTED dance anymore. get rid of a bunch of comments that talk about
how we're stealing from the swapper, which is a lie now.

tweaks from oga@ scrutiny by matthew@
ok miod@ oga@ thib@ beck@

All users of physio(9) now pass NULL as the buf pointer argument, so
no point in keeping it around.

"i like this" thib@ (a while back); ok krw@ and oga@; reminder to
update the man page and tweaks jmc@

Implement vs{,un}lock_device and use it for physio.

Just like normal vs{,un}lock, but in case the pages we get are not dma
accessible, we bounce them, if they are dma acessible, the functions behave
exactly like normal vslock. The plan for the future is to have fault_wire
allocate dma acessible pages so that we don't need to bounce (especially
in cases where the same buffer is reused for physio over and over again),
but for now, keep it as simple as possible.

No need to do the splbio/splx dance around the pool_get in
getphysbuf() as the bufpool is already protected.

From Matthew Dempsky <matthew at dempsky dot org>
OK blambert@

obvious shrinkage.

ok krw@, miod@.

Dynamic buffer cache. Initial diff from mickey@, okay art@ beck@ toby@
deraadt@ dlg@.

'tranfer' -> 'transfer' in comments.

Kernel stack can be swapped. This means that stuff that's on the stack
should never be referenced outside the context of the process to which
this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
doesn't follow the rules here. Instead of trying to track down all
offenders and fix this hairy situation, it makes much more sense
to not swap kernel stacks.

From art@, tested by many some time ago.

typos; from bret lambert

Don't panic for pathological i/o sizes unless minphys() really is
broken. Eliminate an unneeded variable and potential conversion issues
in SCIOCCOMMAND code before calling physio.

Similar to what NetBSD does.

Fixes cdda2wav vs "Billie Holiday - Songs for Distingue Lovers"
problem noted by Alexandre Ratchov. Tested by Alexandre.

ok marco@ pedro@ deraadt@ mickey@

ansi/deregister.
'go for it' deraadt@

typos from Jonathon Gray;

remove caddr_t casts. it's just silly to cast something when the function
takes a void *. convert uiomove to take a void * as well. ok deraadt@

Remove the advertising clause in the UCB license which Berkeley
rescinded 22 July 1999. Proofed by myself and Theo.

First round of __P removal in sys

branches: 1.18.2;
No need for uvm_useracc here now. vslock catches those problems.

Sync in more uvm changes from NetBSD.
This time we're getting rid of KERN_* and VM_PAGER_* error codes and
use errnos instead.

Remove creds from struct buf, move the creds that nfs need into the nfs node.
While in the area, convert nfs node allocation from malloc to pool and do
some cleanups.
Based on the UBC changes in NetBSD. niklas@ ok.

Use bufpool, not malloc.

Replace inclusion of <vm/foo.h> with the correct <uvm/bar.h> when necessary.
(Look ma, I might have broken the tree)

remove old vm

Appease gcc by not using void pointers in arithmetic operations; art@ ok

KNF

Check for failure now that uvm_vslock can return one. Before this, there was
a risk that we could crash when doing physio/sysctl when the system was
completly out of RAM and swap.

Use PHOLD/PRELE instead of manually fiddling with p_holdcnt.

Bring in some new UVM code from NetBSD (not current).

- Introduce a new type of map that are interrupt safe and never allow faults
in them. mb_map and kmem_map are made intrsafe.
- Add "access protection" to uvm_vslock (to be passed down to uvm_fault and
later to pmap_enter).
- madvise(2) now works.
- various cleanups.

branches: 1.7.2;
Init the buffer dependency list.

s/memset/bzero/; art@ ok

mostly name changes and allocation chnges for uvm

indent properly

from netbsd: check user access rights only if UIO_USERSPACE

From NetBSD: 960217 merge

branches: 1.1.1;
Initial revision

Convert infinite sleeps to {m,t}sleep_nsec(9).

ok visa@

Remove some includes include-what-you-use claims don't
have any direct symbols used. Tested for indirect use by compiling
amd64/i386/sparc64 kernels.

ok tedu@ deraadt@

Replace a plethora of historical protection options with just
PROT_NONE, PROT_READ, PROT_WRITE, and PROT_EXEC from mman.h.
PROT_MASK is introduced as the one true way of extracting those bits.
Remove UVM_ADV_* wrapper, using the standard names.
ok doug guenther kettenis

remove uneeded proc.h includes
ok mpi@ kspillner@

KASSERTMSG(9): New kernel assertion with message

KASSERT() is annoying as it only prints the expression as a string. If you
(developers) want to know a little more information, you have to do:

#ifdef DIAGNOSTIC
if (bad)
panic(...);
#endif

KASSERTMSG() replaces it into a single line:

KASSERTMSG(!bad, ...);

Taken from NetBSD.

(There is a concern that KASSERT() messages are too long; consume more memory,
and not friendly for small monitors. This have to be considered & revisited
later.)

"Like" from henning@
Man page review & advices from jmc@ and schwarze@

Revert vnd(4) to its older less strict behavior.

Verified by deraadt@ to fix distrib/sgi/iso.

set flags correctly in physio - now with actual clearing of B_DONE and B_ERROR
between re-use of the same buffer to re-call strategy() - large dd's now
work again which were broken by my last commit of this.
ok krw@

back out last change; it caused dd on usb drive to lock the machine.

ok oga@

stylistic clean of buffer flags assignment to be more obviously not a
destructive assignment of flags.
ok oga@

If uvm_vslock_device() fails, just exit the loop. None of the
after_unlock code is needed if this happens, and running it was even
wrong because we weren't setting b_resid, so uio_offset (and higher
up, f_offset) would be incorrectly adjusted.

Discussed with deraadt@.

Enforce that raw disk I/O is only initiated on block boundaries in
physio().

ok deraadt@

because bufs now come out of a pool rather than a global list of staticly
allocated structs, they wont be shared. we dont have to do the silly
B_WANTED dance anymore. get rid of a bunch of comments that talk about
how we're stealing from the swapper, which is a lie now.

tweaks from oga@ scrutiny by matthew@
ok miod@ oga@ thib@ beck@

All users of physio(9) now pass NULL as the buf pointer argument, so
no point in keeping it around.

"i like this" thib@ (a while back); ok krw@ and oga@; reminder to
update the man page and tweaks jmc@

Implement vs{,un}lock_device and use it for physio.

Just like normal vs{,un}lock, but in case the pages we get are not dma
accessible, we bounce them, if they are dma acessible, the functions behave
exactly like normal vslock. The plan for the future is to have fault_wire
allocate dma acessible pages so that we don't need to bounce (especially
in cases where the same buffer is reused for physio over and over again),
but for now, keep it as simple as possible.

No need to do the splbio/splx dance around the pool_get in
getphysbuf() as the bufpool is already protected.

From Matthew Dempsky <matthew at dempsky dot org>
OK blambert@

obvious shrinkage.

ok krw@, miod@.

Dynamic buffer cache. Initial diff from mickey@, okay art@ beck@ toby@
deraadt@ dlg@.

'tranfer' -> 'transfer' in comments.

Kernel stack can be swapped. This means that stuff that's on the stack
should never be referenced outside the context of the process to which
this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
doesn't follow the rules here. Instead of trying to track down all
offenders and fix this hairy situation, it makes much more sense
to not swap kernel stacks.

From art@, tested by many some time ago.

typos; from bret lambert

Don't panic for pathological i/o sizes unless minphys() really is
broken. Eliminate an unneeded variable and potential conversion issues
in SCIOCCOMMAND code before calling physio.

Similar to what NetBSD does.

Fixes cdda2wav vs "Billie Holiday - Songs for Distingue Lovers"
problem noted by Alexandre Ratchov. Tested by Alexandre.

ok marco@ pedro@ deraadt@ mickey@

ansi/deregister.
'go for it' deraadt@

typos from Jonathon Gray;

remove caddr_t casts. it's just silly to cast something when the function
takes a void *. convert uiomove to take a void * as well. ok deraadt@

Remove the advertising clause in the UCB license which Berkeley
rescinded 22 July 1999. Proofed by myself and Theo.

First round of __P removal in sys

branches: 1.18.2;
No need for uvm_useracc here now. vslock catches those problems.

Sync in more uvm changes from NetBSD.
This time we're getting rid of KERN_* and VM_PAGER_* error codes and
use errnos instead.

Remove creds from struct buf, move the creds that nfs need into the nfs node.
While in the area, convert nfs node allocation from malloc to pool and do
some cleanups.
Based on the UBC changes in NetBSD. niklas@ ok.

Use bufpool, not malloc.

Replace inclusion of <vm/foo.h> with the correct <uvm/bar.h> when necessary.
(Look ma, I might have broken the tree)

remove old vm

Appease gcc by not using void pointers in arithmetic operations; art@ ok

KNF

Check for failure now that uvm_vslock can return one. Before this, there was
a risk that we could crash when doing physio/sysctl when the system was
completly out of RAM and swap.

Use PHOLD/PRELE instead of manually fiddling with p_holdcnt.

Bring in some new UVM code from NetBSD (not current).

- Introduce a new type of map that are interrupt safe and never allow faults
in them. mb_map and kmem_map are made intrsafe.
- Add "access protection" to uvm_vslock (to be passed down to uvm_fault and
later to pmap_enter).
- madvise(2) now works.
- various cleanups.

branches: 1.7.2;
Init the buffer dependency list.

s/memset/bzero/; art@ ok

mostly name changes and allocation chnges for uvm

indent properly

from netbsd: check user access rights only if UIO_USERSPACE

From NetBSD: 960217 merge

branches: 1.1.1;
Initial revision

Remove some includes include-what-you-use claims don't
have any direct symbols used. Tested for indirect use by compiling
amd64/i386/sparc64 kernels.

ok tedu@ deraadt@

Replace a plethora of historical protection options with just
PROT_NONE, PROT_READ, PROT_WRITE, and PROT_EXEC from mman.h.
PROT_MASK is introduced as the one true way of extracting those bits.
Remove UVM_ADV_* wrapper, using the standard names.
ok doug guenther kettenis

remove uneeded proc.h includes
ok mpi@ kspillner@

KASSERTMSG(9): New kernel assertion with message

KASSERT() is annoying as it only prints the expression as a string. If you
(developers) want to know a little more information, you have to do:

#ifdef DIAGNOSTIC
if (bad)
panic(...);
#endif

KASSERTMSG() replaces it into a single line:

KASSERTMSG(!bad, ...);

Taken from NetBSD.

(There is a concern that KASSERT() messages are too long; consume more memory,
and not friendly for small monitors. This have to be considered & revisited
later.)

"Like" from henning@
Man page review & advices from jmc@ and schwarze@

Revert vnd(4) to its older less strict behavior.

Verified by deraadt@ to fix distrib/sgi/iso.

set flags correctly in physio - now with actual clearing of B_DONE and B_ERROR
between re-use of the same buffer to re-call strategy() - large dd's now
work again which were broken by my last commit of this.
ok krw@

back out last change; it caused dd on usb drive to lock the machine.

ok oga@

stylistic clean of buffer flags assignment to be more obviously not a
destructive assignment of flags.
ok oga@

If uvm_vslock_device() fails, just exit the loop. None of the
after_unlock code is needed if this happens, and running it was even
wrong because we weren't setting b_resid, so uio_offset (and higher
up, f_offset) would be incorrectly adjusted.

Discussed with deraadt@.

Enforce that raw disk I/O is only initiated on block boundaries in
physio().

ok deraadt@

because bufs now come out of a pool rather than a global list of staticly
allocated structs, they wont be shared. we dont have to do the silly
B_WANTED dance anymore. get rid of a bunch of comments that talk about
how we're stealing from the swapper, which is a lie now.

tweaks from oga@ scrutiny by matthew@
ok miod@ oga@ thib@ beck@

All users of physio(9) now pass NULL as the buf pointer argument, so
no point in keeping it around.

"i like this" thib@ (a while back); ok krw@ and oga@; reminder to
update the man page and tweaks jmc@

Implement vs{,un}lock_device and use it for physio.

Just like normal vs{,un}lock, but in case the pages we get are not dma
accessible, we bounce them, if they are dma acessible, the functions behave
exactly like normal vslock. The plan for the future is to have fault_wire
allocate dma acessible pages so that we don't need to bounce (especially
in cases where the same buffer is reused for physio over and over again),
but for now, keep it as simple as possible.

No need to do the splbio/splx dance around the pool_get in
getphysbuf() as the bufpool is already protected.

From Matthew Dempsky <matthew at dempsky dot org>
OK blambert@

obvious shrinkage.

ok krw@, miod@.

Dynamic buffer cache. Initial diff from mickey@, okay art@ beck@ toby@
deraadt@ dlg@.

'tranfer' -> 'transfer' in comments.

Kernel stack can be swapped. This means that stuff that's on the stack
should never be referenced outside the context of the process to which
this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
doesn't follow the rules here. Instead of trying to track down all
offenders and fix this hairy situation, it makes much more sense
to not swap kernel stacks.

From art@, tested by many some time ago.

typos; from bret lambert

Don't panic for pathological i/o sizes unless minphys() really is
broken. Eliminate an unneeded variable and potential conversion issues
in SCIOCCOMMAND code before calling physio.

Similar to what NetBSD does.

Fixes cdda2wav vs "Billie Holiday - Songs for Distingue Lovers"
problem noted by Alexandre Ratchov. Tested by Alexandre.

ok marco@ pedro@ deraadt@ mickey@

ansi/deregister.
'go for it' deraadt@

typos from Jonathon Gray;

remove caddr_t casts. it's just silly to cast something when the function
takes a void *. convert uiomove to take a void * as well. ok deraadt@

Remove the advertising clause in the UCB license which Berkeley
rescinded 22 July 1999. Proofed by myself and Theo.

First round of __P removal in sys

branches: 1.18.2;
No need for uvm_useracc here now. vslock catches those problems.

Sync in more uvm changes from NetBSD.
This time we're getting rid of KERN_* and VM_PAGER_* error codes and
use errnos instead.

Remove creds from struct buf, move the creds that nfs need into the nfs node.
While in the area, convert nfs node allocation from malloc to pool and do
some cleanups.
Based on the UBC changes in NetBSD. niklas@ ok.

Use bufpool, not malloc.

Replace inclusion of <vm/foo.h> with the correct <uvm/bar.h> when necessary.
(Look ma, I might have broken the tree)

remove old vm

Appease gcc by not using void pointers in arithmetic operations; art@ ok

KNF

Check for failure now that uvm_vslock can return one. Before this, there was
a risk that we could crash when doing physio/sysctl when the system was
completly out of RAM and swap.

Use PHOLD/PRELE instead of manually fiddling with p_holdcnt.

Bring in some new UVM code from NetBSD (not current).

- Introduce a new type of map that are interrupt safe and never allow faults
in them. mb_map and kmem_map are made intrsafe.
- Add "access protection" to uvm_vslock (to be passed down to uvm_fault and
later to pmap_enter).
- madvise(2) now works.
- various cleanups.

branches: 1.7.2;
Init the buffer dependency list.

s/memset/bzero/; art@ ok

mostly name changes and allocation chnges for uvm

indent properly

from netbsd: check user access rights only if UIO_USERSPACE

From NetBSD: 960217 merge

branches: 1.1.1;
Initial revision