date:20110115

svn commit: r217439 - stable/8/sbin/hastd

2011-01-15 Thread Pawel Jakub Dawidek

Author: pjd
Date: Sat Jan 15 08:18:58 2011
New Revision: 217439
URL: http://svn.freebsd.org/changeset/base/217439

Log:
  MFC r217307,r217308,r217312:
  
  r217307:
  
  Install default signal handlers before masking signals we want to handle.
  It is possible that the parent process ignores some of them and sigtimedwait()
  will never see them, eventhough they are masked.
  
  The most common situation for this to happen is boot process where init(8)
  ignores SIGHUP before starting to execute /etc/rc. This in turn caused
  hastd(8) to ignore SIGHUP.
  
  Reported by:  trasz
  Obtained from:Wheel Systems Sp. z o.o. http://www.wheelsystems.com
  
  r217308:
  
  Add a note that when custom signal handler is installed for a signal,
  signal action is restored to default in child after fork(2).
  In this case there is no need to do anything with dummy SIGCHLD handler,
  because after fork(2) it will be automatically reverted to SIG_IGN.
  
  Obtained from:Wheel Systems Sp. z o.o. http://www.wheelsystems.com
  
  r217312:
  
  execve(2), not fork(2) resets signal handler to the default value (if it isn't
  ignored). Correct comment talking about that.
  
  Pointed out by:   kib

Modified:
  stable/8/sbin/hastd/hastd.c
  stable/8/sbin/hastd/hooks.c
Directory Properties:
  stable/8/sbin/hastd/   (props changed)

Modified: stable/8/sbin/hastd/hastd.c
==
--- stable/8/sbin/hastd/hastd.c Sat Jan 15 03:06:03 2011(r217438)
+++ stable/8/sbin/hastd/hastd.c Sat Jan 15 08:18:58 2011(r217439)
@@ -754,10 +754,18 @@ main(int argc, char *argv[])
assert(cfg != NULL);
 
/*
+* Restore default actions for interesting signals in case parent
+* process (like init(8)) decided to ignore some of them (like SIGHUP).
+*/
+   PJDLOG_VERIFY(signal(SIGHUP, SIG_DFL) != SIG_ERR);
+   PJDLOG_VERIFY(signal(SIGINT, SIG_DFL) != SIG_ERR);
+   PJDLOG_VERIFY(signal(SIGTERM, SIG_DFL) != SIG_ERR);
+   /*
 * Because SIGCHLD is ignored by default, setup dummy handler for it,
 * so we can mask it.
 */
PJDLOG_VERIFY(signal(SIGCHLD, dummy_sighandler) != SIG_ERR);
+
PJDLOG_VERIFY(sigemptyset(&mask) == 0);
PJDLOG_VERIFY(sigaddset(&mask, SIGHUP) == 0);
PJDLOG_VERIFY(sigaddset(&mask, SIGINT) == 0);

Modified: stable/8/sbin/hastd/hooks.c
==
--- stable/8/sbin/hastd/hooks.c Sat Jan 15 03:06:03 2011(r217438)
+++ stable/8/sbin/hastd/hooks.c Sat Jan 15 08:18:58 2011(r217439)
@@ -372,6 +372,11 @@ hook_execv(const char *path, va_list ap)
descriptors();
PJDLOG_VERIFY(sigemptyset(&mask) == 0);
PJDLOG_VERIFY(sigprocmask(SIG_SETMASK, &mask, NULL) == 0);
+   /*
+* Dummy handler set for SIGCHLD in the parent will be restored
+* to SIG_IGN on execv(3) below, so there is no need to do
+* anything with it.
+*/
execv(path, args);
pjdlog_errno(LOG_ERR, "Unable to execute %s", path);
exit(EX_SOFTWARE);
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217440 - stable/8/lib/libc/net

2011-01-15 Thread Konstantin Belousov

Author: kib
Date: Sat Jan 15 08:35:41 2011
New Revision: 217440
URL: http://svn.freebsd.org/changeset/base/217440

Log:
  MFC r217143:
  Fix struct FILE * leak on error (in disabled by default hesiod support code).
  
  PR:   153756

Modified:
  stable/8/lib/libc/net/hesiod.c
Directory Properties:
  stable/8/lib/libc/   (props changed)
  stable/8/lib/libc/locale/   (props changed)
  stable/8/lib/libc/stdtime/   (props changed)
  stable/8/lib/libc/sys/   (props changed)

Modified: stable/8/lib/libc/net/hesiod.c
==
--- stable/8/lib/libc/net/hesiod.c  Sat Jan 15 08:18:58 2011
(r217439)
+++ stable/8/lib/libc/net/hesiod.c  Sat Jan 15 08:35:41 2011
(r217440)
@@ -324,6 +324,7 @@ read_config_file(ctx, filename)
? &ctx->lhs : &ctx->rhs;
*which = strdup(data);
if (!*which) {
+   fclose(fp);
errno = ENOMEM;
return -1;
}
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217441 - stable/8/lib/libc/stdlib

2011-01-15 Thread Konstantin Belousov

Author: kib
Date: Sat Jan 15 08:37:33 2011
New Revision: 217441
URL: http://svn.freebsd.org/changeset/base/217441

Log:
  MFC r217144:
  Fix some style(9) issues.
  Do not use strlcpy() where simple assignment is enough.

Modified:
  stable/8/lib/libc/stdlib/realpath.c
Directory Properties:
  stable/8/lib/libc/   (props changed)
  stable/8/lib/libc/locale/   (props changed)
  stable/8/lib/libc/stdtime/   (props changed)
  stable/8/lib/libc/sys/   (props changed)

Modified: stable/8/lib/libc/stdlib/realpath.c
==
--- stable/8/lib/libc/stdlib/realpath.c Sat Jan 15 08:35:41 2011
(r217440)
+++ stable/8/lib/libc/stdlib/realpath.c Sat Jan 15 08:37:33 2011
(r217441)
@@ -54,7 +54,7 @@ realpath(const char * __restrict path, c
char *p, *q, *s;
size_t left_len, resolved_len;
unsigned symlinks;
-   int serrno, slen, m;
+   int m, serrno, slen;
char left[PATH_MAX], next_token[PATH_MAX], symlink[PATH_MAX];
 
if (path == NULL) {
@@ -73,7 +73,6 @@ realpath(const char * __restrict path, c
m = 1;
} else
m = 0;
-
symlinks = 0;
if (path[0] == '/') {
resolved[0] = '/';
@@ -86,8 +85,10 @@ realpath(const char * __restrict path, c
if (getcwd(resolved, PATH_MAX) == NULL) {
if (m)
free(resolved);
-   else
-   strlcpy(resolved, ".", PATH_MAX);
+   else {
+   resolved[0] = '.';
+   resolved[1] = '\0';
+   }
return (NULL);
}
resolved_len = strlen(resolved);
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

Re: svn commit: r217369 - in head/sys: cam/scsi sys

2011-01-15 Thread Garrett Cooper

On Fri, Jan 14, 2011 at 10:27 PM, Bruce Evans  wrote:
> On Fri, 14 Jan 2011, Garrett Cooper wrote:
>
>> On Fri, Jan 14, 2011 at 6:42 PM, Bruce Evans  wrote:
>>>
>>> On Fri, 14 Jan 2011 m...@freebsd.org wrote:
>>>
 On Thu, Jan 13, 2011 at 10:50 PM, Bruce Evans 
 wrote:
>
> On Thu, 13 Jan 2011 m...@freebsd.org wrote:
>
>> There appear to be 330 uses of SYSCTL and QUAD on the same line in
>> CURRENT.  This seems reasonable to change them to S64, U64 and X64 so
>> they correctly reflect the size they operate upon.
>>
>> What do y'all think?
>
> Now I suggest delaying this until they can be renamed to a type-
> generic
> SYSCTL_INT() (would probably need to be spelled differently, SYSCTL_I()
> say, even if SYSCTL_INT() was changed at the same time).

 I'm torn on this one.  The compiler knows the type (unless, for
 SYSCTL_INT, NULL/0 is used, but that is also a compile-time check),
 but to interpret it requires the use of __builtin_foo which is a gcc
 extension and not part of standard C.

 Philosophically, while I like this kind of letting the compiler do the
 work, if you want C++ you know where to find it.
>>>
>>> Oops.  I think sizeof() and issigned() can be used to determine the type
>>> well enough in functions and initialized data (do a fuller type check if
>>> the compiler supports it), but I don't know how to do this in static
>>> sysctl declarations (since sizeof() can't be used in cpp expressions).
>>
>>   Why not just create some dumb testcases that can be run at build
>> time to determine that for you?
>
> Well, how?  You are given SYSCTL_I(&var, ...) and have to convert this
> to what is now in SYSCTL_INT(), using only the type of var, in hundreds
> or thousands of files.  I don't even know how to do this with a test
> case for each file, short of parsing all the files.  Oops, I do know
> how to translate from sizeof(var) to CTLTYPE_INT or CTLTYPE_UINT.
> That's just (sizeof(var) == sizeof(int) ? CTLTYPE_INT : ...).  The
> signness is harder (might need gnu typeof(), but not the recent type
> checking attributes).  This won't convert from SYSCTL_I() existing
> SYSCTL_INT() (the switch on the size would have to be in an ifdef for
> that, but sizeof() doesn't work in ifdefs), but it works for generating
> CTLTYPE_* internally SYSCTL_I().  The difficulty is converting from a
> bare variable `var' to an integer representing the signedness of its
> type, without using an unportability like typeof().  With typeof(), this
> is:
>
>        /* Only works for arithmetic types: */
>        #define isinteger(var)  ((typeof(var))0.1 == 0)
>        #define issigned(var)   ((typeof(var))-1 < 0)
>        ...

This is what I meant:

$ cat test_warnings.c
#include 

size_t x = (int) -1;
int y = 200L;
$ gcc -Wconversion -Wstrict-overflow -Wsign-compare -c test_warnings.c
test_size_t.c:3: warning: negative integer implicitly converted to unsigned type
test_size_t.c:4: warning: overflow in implicit constant conversion
$

With the right CFLAGS and a few properly written tests, and a few
make rules, you can figure out what's what pretty easily *shrugs*.
Thanks,
-Garrett
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217442 - stable/8/sys/kern

2011-01-15 Thread Konstantin Belousov

Author: kib
Date: Sat Jan 15 09:25:19 2011
New Revision: 217442
URL: http://svn.freebsd.org/changeset/base/217442

Log:
  MFC r217150:
  Collect code to translate between vm_prot_t and p_flags into helper
  functions.

Modified:
  stable/8/sys/kern/imgact_elf.c
Directory Properties:
  stable/8/sys/   (props changed)
  stable/8/sys/amd64/include/xen/   (props changed)
  stable/8/sys/cddl/contrib/opensolaris/   (props changed)
  stable/8/sys/contrib/dev/acpica/   (props changed)
  stable/8/sys/contrib/pf/   (props changed)

Modified: stable/8/sys/kern/imgact_elf.c
==
--- stable/8/sys/kern/imgact_elf.c  Sat Jan 15 08:37:33 2011
(r217441)
+++ stable/8/sys/kern/imgact_elf.c  Sat Jan 15 09:25:19 2011
(r217442)
@@ -86,6 +86,8 @@ static boolean_t __elfN(freebsd_trans_os
 static boolean_t kfreebsd_trans_osrel(const Elf_Note *note, int32_t *osrel);
 static boolean_t __elfN(check_note)(struct image_params *imgp,
 Elf_Brandnote *checknote, int32_t *osrel);
+static vm_prot_t __elfN(trans_prot)(Elf_Word);
+static Elf_Word __elfN(untrans_prot)(vm_prot_t);
 
 SYSCTL_NODE(_kern, OID_AUTO, __CONCAT(elf, __ELF_WORD_SIZE), CTLFLAG_RW, 0,
 "");
@@ -632,14 +634,7 @@ __elfN(load_file)(struct proc *p, const 
for (i = 0, numsegs = 0; i < hdr->e_phnum; i++) {
if (phdr[i].p_type == PT_LOAD && phdr[i].p_memsz != 0) {
/* Loadable segment */
-   prot = 0;
-   if (phdr[i].p_flags & PF_X)
-   prot |= VM_PROT_EXECUTE;
-   if (phdr[i].p_flags & PF_W)
-   prot |= VM_PROT_WRITE;
-   if (phdr[i].p_flags & PF_R)
-   prot |= VM_PROT_READ;
-
+   prot = __elfN(trans_prot)(phdr[i].p_flags);
if ((error = __elfN(load_section)(vmspace,
imgp->object, phdr[i].p_offset,
(caddr_t)(uintptr_t)phdr[i].p_vaddr + rbase,
@@ -780,13 +775,7 @@ __CONCAT(exec_, __elfN(imgact))(struct i
case PT_LOAD:   /* Loadable segment */
if (phdr[i].p_memsz == 0)
break;
-   prot = 0;
-   if (phdr[i].p_flags & PF_X)
-   prot |= VM_PROT_EXECUTE;
-   if (phdr[i].p_flags & PF_W)
-   prot |= VM_PROT_WRITE;
-   if (phdr[i].p_flags & PF_R)
-   prot |= VM_PROT_READ;
+   prot = __elfN(trans_prot)(phdr[i].p_flags);
 
 #if defined(__ia64__) && __ELF_WORD_SIZE == 32 && defined(IA32_ME_HARDER)
/*
@@ -1086,13 +1075,7 @@ cb_put_phdr(entry, closure)
phdr->p_paddr = 0;
phdr->p_filesz = phdr->p_memsz = entry->end - entry->start;
phdr->p_align = PAGE_SIZE;
-   phdr->p_flags = 0;
-   if (entry->protection & VM_PROT_READ)
-   phdr->p_flags |= PF_R;
-   if (entry->protection & VM_PROT_WRITE)
-   phdr->p_flags |= PF_W;
-   if (entry->protection & VM_PROT_EXECUTE)
-   phdr->p_flags |= PF_X;
+   phdr->p_flags = __elfN(untrans_prot)(entry->protection);
 
phc->offset += phdr->p_filesz;
phc->phdr++;
@@ -1483,3 +1466,33 @@ static struct execsw __elfN(execsw) = {
__XSTRING(__CONCAT(ELF, __ELF_WORD_SIZE))
 };
 EXEC_SET(__CONCAT(elf, __ELF_WORD_SIZE), __elfN(execsw));
+
+static vm_prot_t
+__elfN(trans_prot)(Elf_Word flags)
+{
+   vm_prot_t prot;
+
+   prot = 0;
+   if (flags & PF_X)
+   prot |= VM_PROT_EXECUTE;
+   if (flags & PF_W)
+   prot |= VM_PROT_WRITE;
+   if (flags & PF_R)
+   prot |= VM_PROT_READ;
+   return (prot);
+}
+
+static Elf_Word
+__elfN(untrans_prot)(vm_prot_t prot)
+{
+   Elf_Word flags;
+
+   flags = 0;
+   if (prot & VM_PROT_EXECUTE)
+   flags |= PF_X;
+   if (prot & VM_PROT_READ)
+   flags |= PF_R;
+   if (prot & VM_PROT_WRITE)
+   flags |= PF_W;
+   return (flags);
+}
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217443 - stable/7/sys/geom

2011-01-15 Thread Jaakko Heinonen

Author: jh
Date: Sat Jan 15 09:36:31 2011
New Revision: 217443
URL: http://svn.freebsd.org/changeset/base/217443

Log:
  MFC r214748:
  
  Extend the g_eventlock mutex coverage in one_event() to include setting
  of the EV_DONE flag and use the mutex to protect against losing wakeups
  in g_waitfor_event().

Modified:
  stable/7/sys/geom/geom_event.c
Directory Properties:
  stable/7/sys/   (props changed)
  stable/7/sys/cddl/contrib/opensolaris/   (props changed)
  stable/7/sys/contrib/dev/acpica/   (props changed)
  stable/7/sys/contrib/pf/   (props changed)

Modified: stable/7/sys/geom/geom_event.c
==
--- stable/7/sys/geom/geom_event.c  Sat Jan 15 09:25:19 2011
(r217442)
+++ stable/7/sys/geom/geom_event.c  Sat Jan 15 09:36:31 2011
(r217443)
@@ -212,11 +212,12 @@ one_event(void)
g_topology_assert();
mtx_lock(&g_eventlock);
TAILQ_REMOVE(&g_events, ep, events);
-   mtx_unlock(&g_eventlock);
if (ep->flag & EV_WAKEUP) {
ep->flag |= EV_DONE;
+   mtx_unlock(&g_eventlock);
wakeup(ep);
} else {
+   mtx_unlock(&g_eventlock);
g_free(ep);
}
g_topology_unlock();
@@ -355,11 +356,14 @@ g_waitfor_event(g_event_t *func, void *a
va_end(ap);
if (error)
return (error);
-   do 
-   tsleep(ep, PRIBIO, "g_waitfor_event", hz);
-   while (!(ep->flag & EV_DONE));
+
+   mtx_lock(&g_eventlock);
+   while (!(ep->flag & EV_DONE))
+   msleep(ep, &g_eventlock, PRIBIO, "g_waitfor_event", hz);
if (ep->flag & EV_CANCELED)
error = EAGAIN;
+   mtx_unlock(&g_eventlock);
+
g_free(ep);
return (error);
 }
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217444 - head/sys/cam/ata

2011-01-15 Thread Alexander Motin

Author: mav
Date: Sat Jan 15 09:43:25 2011
New Revision: 217444
URL: http://svn.freebsd.org/changeset/base/217444

Log:
  Some old WD SATA disks report supported and enabled device-initiated
  interface power management, but return ABORT error on attempt to disable
  it. Make CAM SATA probe sequence ignore this error, as it is not fatal.

Modified:
  head/sys/cam/ata/ata_xpt.c

Modified: head/sys/cam/ata/ata_xpt.c
==
--- head/sys/cam/ata/ata_xpt.c  Sat Jan 15 09:36:31 2011(r217443)
+++ head/sys/cam/ata/ata_xpt.c  Sat Jan 15 09:43:25 2011(r217444)
@@ -727,6 +727,7 @@ probedone(struct cam_periph *periph, uni
struct ata_params *ident_buf;
probe_softc *softc;
struct cam_path *path;
+   cam_status status;
u_int32_t  priority;
u_int caps;
int found = 1;
@@ -751,6 +752,7 @@ probedone(struct cam_periph *periph, uni
xpt_release_devq(done_ccb->ccb_h.path, /*count*/1,
 /*run_queue*/TRUE);
}
+   status = done_ccb->ccb_h.status & CAM_STATUS_MASK;
if (softc->restart) {
softc->faults++;
if ((done_ccb->ccb_h.status & CAM_STATUS_MASK) ==
@@ -760,12 +762,24 @@ probedone(struct cam_periph *periph, uni
goto done;
else
softc->restart = 0;
-   } else
+
/* Old PIO2 devices may not support mode setting. */
-   if (softc->action == PROBE_SETMODE &&
+   } else if (softc->action == PROBE_SETMODE &&
+   status == CAM_ATA_STATUS_ERROR &&
ata_max_pmode(ident_buf) <= ATA_PIO2 &&
-   (ident_buf->capabilities1 & ATA_SUPPORT_IORDY) == 0)
+   (ident_buf->capabilities1 & ATA_SUPPORT_IORDY) == 0) {
+   goto noerror;
+
+   /*
+* Some old WD SATA disks report supported and enabled
+* device-initiated interface power management, but return
+* ABORT on attempt to disable it.
+*/
+   } else if (softc->action == PROBE_SETPM &&
+   status == CAM_ATA_STATUS_ERROR) {
goto noerror;
+   }
+
/*
 * If we get to this point, we got an error status back
 * from the inquiry and the error status doesn't require
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

Re: svn commit: r217369 - in head/sys: cam/scsi sys

2011-01-15 Thread Bruce Evans


On Sat, 15 Jan 2011, Garrett Cooper wrote:


On Fri, Jan 14, 2011 at 10:27 PM, Bruce Evans  wrote:

On Fri, 14 Jan 2011, Garrett Cooper wrote:


On Fri, Jan 14, 2011 at 6:42 PM, Bruce Evans  wrote:



...
Oops. ?I think sizeof() and issigned() can be used to determine the type
well enough in functions and initialized data (do a fuller type check if
the compiler supports it), but I don't know how to do this in static
sysctl declarations (since sizeof() can't be used in cpp expressions).


? Why not just create some dumb testcases that can be run at build
time to determine that for you?


Well, how? ?You are given SYSCTL_I(&var, ...) and have to convert this
to what is now in SYSCTL_INT(), using only the type of var, in hundreds
or thousands of files. ?I don't even know how to do this with a test
...

? ? ? ?/* Only works for arithmetic types: */
? ? ? ?#define isinteger(var) ?((typeof(var))0.1 == 0)
? ? ? ?#define issigned(var) ? ((typeof(var))-1 < 0)
? ? ? ?...


   This is what I meant:

$ cat test_warnings.c
#include 

size_t x = (int) -1;
int y = 200L;
$ gcc -Wconversion -Wstrict-overflow -Wsign-compare -c test_warnings.c
test_size_t.c:3: warning: negative integer implicitly converted to unsigned type
test_size_t.c:4: warning: overflow in implicit constant conversion
$

   With the right CFLAGS and a few properly written tests, and a few
make rules, you can figure out what's what pretty easily *shrugs*.


That's a lot more parsing than seems reasonable.

Anyway, we already depend on gnu __typeof() being avaible for the much
more central API of pcpu.  All pcpu accesses on amd64 and i386 use it,
except curthread (curthread has been hacked to use a more direct asm
for compile-time efficiency).

SYSCTL_I() works even better that I first thought.  It automatically
gives support for all typedefed integral types.  No SYSCTL_FOO_T()s
with messy MD ifdefs for matching up foo_t with an integral type are
needed.  Instead there are even messier MI ifdefs in SYSCTL_I() :-).
But not so many.  There are hundreds of typedefed types of interest,
but only 8 different integer types to map to (8/16/32/64 bits signed
and unsigned).  The complication that int64_t is plain long on 64
bit arches but long long on 32-bit arches still arises.  CTLTYPE_QUAD
can be associated with int64_t on all arches, but the format for printing
these is arch-dependent.  (Note that quad_t's cannot be printed using
%qd, and %qd is unusable, since %qd is just an alias for %lld, while
quad_t is an alias for int64_t and int64_t is plain long on 64-bit
arches so its format is %ld on these arches and %lld on others.)

Bruce___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

Re: svn commit: r217369 - in head/sys: cam/scsi sys

2011-01-15 Thread mdf

On Sat, Jan 15, 2011 at 6:55 AM, Bruce Evans  wrote:
> On Sat, 15 Jan 2011, Garrett Cooper wrote:
>
>> On Fri, Jan 14, 2011 at 10:27 PM, Bruce Evans 
>> wrote:
>>>
>>> On Fri, 14 Jan 2011, Garrett Cooper wrote:
>>>
 On Fri, Jan 14, 2011 at 6:42 PM, Bruce Evans 
 wrote:
>
> ...
> Oops.  I think sizeof() and issigned() can be used to determine the
> type
> well enough in functions and initialized data (do a fuller type check
> if
> the compiler supports it), but I don't know how to do this in static
> sysctl declarations (since sizeof() can't be used in cpp expressions).

   Why not just create some dumb testcases that can be run at build
 time to determine that for you?
>>>
>>> Well, how?  You are given SYSCTL_I(&var, ...) and have to convert this
>>> to what is now in SYSCTL_INT(), using only the type of var, in hundreds
>>> or thousands of files.  I don't even know how to do this with a test
>>> ...
>>>
>>>        /* Only works for arithmetic types: */
>>>        #define isinteger(var)  ((typeof(var))0.1 == 0)
>>>        #define issigned(var)   ((typeof(var))-1 < 0)
>>>        ...
>>
>>   This is what I meant:
>>
>> $ cat test_warnings.c
>> #include 
>>
>> size_t x = (int) -1;
>> int y = 200L;
>> $ gcc -Wconversion -Wstrict-overflow -Wsign-compare -c test_warnings.c
>> test_size_t.c:3: warning: negative integer implicitly converted to
>> unsigned type
>> test_size_t.c:4: warning: overflow in implicit constant conversion
>> $
>>
>>   With the right CFLAGS and a few properly written tests, and a few
>> make rules, you can figure out what's what pretty easily *shrugs*.
>
> That's a lot more parsing than seems reasonable.
>
> Anyway, we already depend on gnu __typeof() being avaible for the much
> more central API of pcpu.  All pcpu accesses on amd64 and i386 use it,
> except curthread (curthread has been hacked to use a more direct asm
> for compile-time efficiency).
>
> SYSCTL_I() works even better that I first thought.  It automatically
> gives support for all typedefed integral types.  No SYSCTL_FOO_T()s
> with messy MD ifdefs for matching up foo_t with an integral type are
> needed.  Instead there are even messier MI ifdefs in SYSCTL_I() :-).
> But not so many.  There are hundreds of typedefed types of interest,
> but only 8 different integer types to map to (8/16/32/64 bits signed
> and unsigned).  The complication that int64_t is plain long on 64
> bit arches but long long on 32-bit arches still arises.  CTLTYPE_QUAD
> can be associated with int64_t on all arches, but the format for printing
> these is arch-dependent.  (Note that quad_t's cannot be printed using
> %qd, and %qd is unusable, since %qd is just an alias for %lld, while
> quad_t is an alias for int64_t and int64_t is plain long on 64-bit
> arches so its format is %ld on these arches and %lld on others.)

The printing is done entirely in user-space, so it's not too bad.  I
had figured to upcast everything to [u]intmax_t anyways, and what to
cast from would just be [u]int32_t and [u]int64_t depending on
reported size.  Anything smaller should be copyout(9)'d as a 4 bit
quantity.

But, there's two factors at play here.  The first is sysctl(8) which
asks for the size when reading and manages it.  The second is
sysctl(2) where we have code like SCTL_MASK32 so a long on a 64-bit
kernel looks like a long to a 32-bit app.  It would be simpler to
expose this as an 8-byte quantity on 64-bit kernel, and sysctl(8) will
deal with this just fine, but that would break some uses of sysctl(2).
 However, I *think* it is safe to check the sysctl_req's oldidx/oldlen
to see what size the user-space expects and cast a kernel long to int
before SYSCTL_OUT.

$WORK has gotten busy so I haven't had a chance to work on this much
but I'm hopeful that the missus will watch the kids this weekend and
allow me to hack.

Thanks,
matthew
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217445 - stable/8/sys/net

2011-01-15 Thread Robert Watson

Author: rwatson
Date: Sat Jan 15 18:41:42 2011
New Revision: 217445
URL: http://svn.freebsd.org/changeset/base/217445

Log:
  Merge r204199 from head to stable/8:
  
Export netisr configuration and statistics to userspace via sysctl(9).
  
Sponsored by:   Juniper Networks

Modified:
  stable/8/sys/net/netisr.c
  stable/8/sys/net/netisr.h
Directory Properties:
  stable/8/sys/   (props changed)
  stable/8/sys/amd64/include/xen/   (props changed)
  stable/8/sys/cddl/contrib/opensolaris/   (props changed)
  stable/8/sys/contrib/dev/acpica/   (props changed)
  stable/8/sys/contrib/pf/   (props changed)

Modified: stable/8/sys/net/netisr.c
==
--- stable/8/sys/net/netisr.c   Sat Jan 15 09:43:25 2011(r217444)
+++ stable/8/sys/net/netisr.c   Sat Jan 15 18:41:42 2011(r217445)
@@ -1,7 +1,11 @@
 /*-
  * Copyright (c) 2007-2009 Robert N. M. Watson
+ * Copyright (c) 2010 Juniper Networks, Inc.
  * All rights reserved.
  *
+ * This software was developed by Robert N. M. Watson under contract
+ * to Juniper Networks, Inc.
+ *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions
  * are met:
@@ -1118,6 +1122,170 @@ netisr_start(void *arg)
 }
 SYSINIT(netisr_start, SI_SUB_SMP, SI_ORDER_MIDDLE, netisr_start, NULL);
 
+/*
+ * Sysctl monitoring for netisr: query a list of registered protocols.
+ */
+static int
+sysctl_netisr_proto(SYSCTL_HANDLER_ARGS)
+{
+   struct rm_priotracker tracker;
+   struct sysctl_netisr_proto *snpp, *snp_array;
+   struct netisr_proto *npp;
+   u_int counter, proto;
+   int error;
+
+   if (req->newptr != NULL)
+   return (EINVAL);
+   snp_array = malloc(sizeof(*snp_array) * NETISR_MAXPROT, M_TEMP,
+   M_ZERO | M_WAITOK);
+   counter = 0;
+   NETISR_RLOCK(&tracker);
+   for (proto = 0; proto < NETISR_MAXPROT; proto++) {
+   npp = &np[proto];
+   if (npp->np_name == NULL)
+   continue;
+   snpp = &snp_array[counter];
+   snpp->snp_version = sizeof(*snpp);
+   strlcpy(snpp->snp_name, npp->np_name, NETISR_NAMEMAXLEN);
+   snpp->snp_proto = proto;
+   snpp->snp_qlimit = npp->np_qlimit;
+   snpp->snp_policy = npp->np_policy;
+   if (npp->np_m2flow != NULL)
+   snpp->snp_flags |= NETISR_SNP_FLAGS_M2FLOW;
+   if (npp->np_m2cpuid != NULL)
+   snpp->snp_flags |= NETISR_SNP_FLAGS_M2CPUID;
+   if (npp->np_drainedcpu != NULL)
+   snpp->snp_flags |= NETISR_SNP_FLAGS_DRAINEDCPU;
+   counter++;
+   }
+   NETISR_RUNLOCK(&tracker);
+   KASSERT(counter < NETISR_MAXPROT,
+   ("sysctl_netisr_proto: counter too big (%d)", counter));
+   error = SYSCTL_OUT(req, snp_array, sizeof(*snp_array) * counter);
+   free(snp_array, M_TEMP);
+   return (error);
+}
+
+SYSCTL_PROC(_net_isr, OID_AUTO, proto,
+CTLFLAG_RD|CTLTYPE_STRUCT|CTLFLAG_MPSAFE, 0, 0, sysctl_netisr_proto,
+"S,sysctl_netisr_proto",
+"Return list of protocols registered with netisr");
+
+/*
+ * Sysctl monitoring for netisr: query a list of workstreams.
+ */
+static int
+sysctl_netisr_workstream(SYSCTL_HANDLER_ARGS)
+{
+   struct rm_priotracker tracker;
+   struct sysctl_netisr_workstream *snwsp, *snws_array;
+   struct netisr_workstream *nwsp;
+   u_int counter, cpuid;
+   int error;
+
+   if (req->newptr != NULL)
+   return (EINVAL);
+   snws_array = malloc(sizeof(*snws_array) * MAXCPU, M_TEMP,
+   M_ZERO | M_WAITOK);
+   counter = 0;
+   NETISR_RLOCK(&tracker);
+   for (cpuid = 0; cpuid < MAXCPU; cpuid++) {
+   if (CPU_ABSENT(cpuid))
+   continue;
+   nwsp = DPCPU_ID_PTR(cpuid, nws);
+   if (nwsp->nws_intr_event == NULL)
+   continue;
+   NWS_LOCK(nwsp);
+   snwsp = &snws_array[counter];
+   snwsp->snws_version = sizeof(*snwsp);
+
+   /*
+* For now, we equate workstream IDs and CPU IDs in the
+* kernel, but expose them independently to userspace in case
+* that assumption changes in the future.
+*/
+   snwsp->snws_wsid = cpuid;
+   snwsp->snws_cpu = cpuid;
+   if (nwsp->nws_intr_event != NULL)
+   snwsp->snws_flags |= NETISR_SNWS_FLAGS_INTR;
+   NWS_UNLOCK(nwsp);
+   counter++;
+   }
+   NETISR_RUNLOCK(&tracker);
+   KASSERT(counter < MAXCPU,
+   ("sysctl_netisr_workstream: counter too big (%d)", counter));
+   error = SYSCTL_OUT(req, snws_array, sizeof(*snws_array) * counter);
+   free(snws_array, M_TEMP);

svn commit: r217446 - stable/8/usr.bin/netstat

2011-01-15 Thread Robert Watson

Author: rwatson
Date: Sat Jan 15 18:51:45 2011
New Revision: 217446
URL: http://svn.freebsd.org/changeset/base/217446

Log:
  Merge r204202 from head to stable/8:
  
Teach netstat(1) to print out netisr statistics when given the -Q argument.
Currently supports only reporting on live systems via sysctl, kmem support
needs to be edded.
  
Sponsored by:   Juniper Networks

Added:
  stable/8/usr.bin/netstat/netisr.c
 - copied unchanged from r204202, head/usr.bin/netstat/netisr.c
Modified:
  stable/8/usr.bin/netstat/Makefile
  stable/8/usr.bin/netstat/main.c
  stable/8/usr.bin/netstat/netstat.1
  stable/8/usr.bin/netstat/netstat.h
Directory Properties:
  stable/8/usr.bin/netstat/   (props changed)

Modified: stable/8/usr.bin/netstat/Makefile
==
--- stable/8/usr.bin/netstat/Makefile   Sat Jan 15 18:41:42 2011
(r217445)
+++ stable/8/usr.bin/netstat/Makefile   Sat Jan 15 18:51:45 2011
(r217446)
@@ -4,7 +4,7 @@
 .include 
 
 PROG=  netstat
-SRCS=  if.c inet.c main.c mbuf.c mroute.c route.c \
+SRCS=  if.c inet.c main.c mbuf.c mroute.c netisr.c route.c \
unix.c atalk.c mroute6.c ipsec.c bpf.c pfkey.c sctp.c
 
 WARNS?=3

Modified: stable/8/usr.bin/netstat/main.c
==
--- stable/8/usr.bin/netstat/main.c Sat Jan 15 18:41:42 2011
(r217445)
+++ stable/8/usr.bin/netstat/main.c Sat Jan 15 18:51:45 2011
(r217446)
@@ -336,6 +336,7 @@ int noutputs = 0;   /* how much outputs be
 intnumeric_addr;   /* show addresses numerically */
 intnumeric_port;   /* show ports numerically */
 static int pflag;  /* show given protocol */
+intQflag;  /* show netisr information */
 intrflag;  /* show routing tables (or routing stats) */
 intsflag;  /* show protocol statistics */
 inttflag;  /* show i/f watchdog timers */
@@ -447,6 +448,9 @@ main(int argc, char *argv[])
}
pflag = 1;
break;
+   case 'Q':
+   Qflag = 1;
+   break;
case 'q':
noutputs = atoi(optarg);
if (noutputs != 0)
@@ -534,6 +538,12 @@ main(int argc, char *argv[])
mbpr(NULL, 0);
exit(0);
}
+   if (Qflag) {
+   if (!live)
+   usage();
+   netisr_stats();
+   exit(0);
+   }
 #if 0
/*
 * Keep file descriptors open to avoid overhead
@@ -789,7 +799,7 @@ name2protox(const char *name)
 static void
 usage(void)
 {
-   (void)fprintf(stderr, "%s\n%s\n%s\n%s\n%s\n%s\n%s\n%s\n%s\n%s\n%s\n",
+   (void)fprintf(stderr, 
"%s\n%s\n%s\n%s\n%s\n%s\n%s\n%s\n%s\n%s\n%s\n%s\n",
 "usage: netstat [-AaLnSTWx] [-f protocol_family | -p protocol]\n"
 "   [-M core] [-N system]",
 "   netstat -i | -I interface [-abdhntW] [-f address_family]\n"
@@ -804,6 +814,7 @@ usage(void)
 "   netstat -r [-AanW] [-f address_family] [-M core] [-N system]",
 "   netstat -rs [-s] [-M core] [-N system]",
 "   netstat -g [-W] [-f address_family] [-M core] [-N system]",
-"   netstat -gs [-s] [-f address_family] [-M core] [-N system]");
+"   netstat -gs [-s] [-f address_family] [-M core] [-N system]",
+"   netstat -Q");
exit(1);
 }

Copied: stable/8/usr.bin/netstat/netisr.c (from r204202, 
head/usr.bin/netstat/netisr.c)
==
--- /dev/null   00:00:00 1970   (empty, because file is newly added)
+++ stable/8/usr.bin/netstat/netisr.c   Sat Jan 15 18:51:45 2011
(r217446, copy of r204202, head/usr.bin/netstat/netisr.c)
@@ -0,0 +1,268 @@
+/*-
+ * Copyright (c) 2010 Juniper Networks, Inc.
+ * All rights reserved.
+ *
+ * This software was developed by Robert N. M. Watson under contract
+ * to Juniper Networks, Inc.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *notice, this list of conditions and the following disclaimer in the
+ *documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ *

svn commit: r217447 - stable/8/share/man/man9

2011-01-15 Thread Robert Watson

Author: rwatson
Date: Sat Jan 15 18:56:51 2011
New Revision: 217447
URL: http://svn.freebsd.org/changeset/base/217447

Log:
  Merge r204206 from head to stable/8:
  
Cross-reference netstat(1) from netisr(9) since you can now use it to
monitor netisr status.
  
Sponsored by:   Juniper Networks

Modified:
  stable/8/share/man/man9/netisr.9
Directory Properties:
  stable/8/share/man/man9/   (props changed)

Modified: stable/8/share/man/man9/netisr.9
==
--- stable/8/share/man/man9/netisr.9Sat Jan 15 18:51:45 2011
(r217446)
+++ stable/8/share/man/man9/netisr.9Sat Jan 15 18:56:51 2011
(r217447)
@@ -27,7 +27,7 @@
 .\"
 .\" $FreeBSD$
 .\"
-.Dd July 26, 2009
+.Dd February 22, 2010
 .Dt NETISR 9
 .Os
 .Sh NAME
@@ -66,6 +66,8 @@ The
 .Nm
 kernel interface suite allows device drivers (and other packet sources) to
 direct packets to protocols for directly dispatched or deferred processing.
+Protocol registration and work stream statistics may be monitored using
+.Xr netstat 1 .
 .Ss Protocol registration
 Protocols register and unregister handlers using
 .Fn netisr_register
@@ -209,6 +211,7 @@ IPv6
 .It Dv NETISR_NATM
 ATM
 .It Dv NETISR_EPAIR
+.Xr netstat 1 ,
 .Xr epair 4
 .El
 .Sh AUTHORS
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217448 - stable/8/usr.bin/netstat

2011-01-15 Thread Robert Watson

Author: rwatson
Date: Sat Jan 15 19:00:22 2011
New Revision: 217448
URL: http://svn.freebsd.org/changeset/base/217448

Log:
  Merge mergeinfo only for r204207 from head to stable/8; the edit itself
  was apparently merged as part of a separate change:
  
Update date on netstat(1) for -Q.
  
Suggested by:   bz

Modified:
Directory Properties:
  stable/8/usr.bin/netstat/   (props changed)
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217449 - stable/8/sys/net

2011-01-15 Thread Robert Watson

Author: rwatson
Date: Sat Jan 15 19:02:08 2011
New Revision: 217449
URL: http://svn.freebsd.org/changeset/base/217449

Log:
  Merge r204208 from head to stable/8:
  
Fix constant assignment for netisr protocol information sysctl.
  
Spotted by: bz

Modified:
  stable/8/sys/net/netisr.h
Directory Properties:
  stable/8/sys/   (props changed)
  stable/8/sys/amd64/include/xen/   (props changed)
  stable/8/sys/cddl/contrib/opensolaris/   (props changed)
  stable/8/sys/contrib/dev/acpica/   (props changed)
  stable/8/sys/contrib/pf/   (props changed)

Modified: stable/8/sys/net/netisr.h
==
--- stable/8/sys/net/netisr.h   Sat Jan 15 19:00:22 2011(r217448)
+++ stable/8/sys/net/netisr.h   Sat Jan 15 19:02:08 2011(r217449)
@@ -92,7 +92,7 @@ struct sysctl_netisr_proto {
  */
 #defineNETISR_SNP_FLAGS_M2FLOW 0x0001  /* nh_m2flow */
 #defineNETISR_SNP_FLAGS_M2CPUID0x0002  /* nh_m2cpuid */
-#defineNETISR_SNP_FLAGS_DRAINEDCPU 0x0003  /* 
nh_drainedcpu */
+#defineNETISR_SNP_FLAGS_DRAINEDCPU 0x0004  /* 
nh_drainedcpu */
 
 /*
  * Next, a structure per-workstream, with per-protocol data, exported as
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217450 - stable/8/sys/net

2011-01-15 Thread Robert Watson

Author: rwatson
Date: Sat Jan 15 19:05:06 2011
New Revision: 217450
URL: http://svn.freebsd.org/changeset/base/217450

Log:
  Merge r204303 from head to stable/8:
  
Fix edge cases in several KASSERTs: use <= rather than < when testing that
counters have not gone about MAXCPU or NETISR_MAXPROT.  These problems
caused panics on UP kernels with INVARIANTS when using sysctl -a, but
would also have caused problems for 32-core boxes or if the netisr
protocol vector was fully populated.
  
Reported by:nwhitehorn, Neel Natu 

Modified:
  stable/8/sys/net/netisr.c
Directory Properties:
  stable/8/sys/   (props changed)
  stable/8/sys/amd64/include/xen/   (props changed)
  stable/8/sys/cddl/contrib/opensolaris/   (props changed)
  stable/8/sys/contrib/dev/acpica/   (props changed)
  stable/8/sys/contrib/pf/   (props changed)

Modified: stable/8/sys/net/netisr.c
==
--- stable/8/sys/net/netisr.c   Sat Jan 15 19:02:08 2011(r217449)
+++ stable/8/sys/net/netisr.c   Sat Jan 15 19:05:06 2011(r217450)
@@ -1159,7 +1159,7 @@ sysctl_netisr_proto(SYSCTL_HANDLER_ARGS)
counter++;
}
NETISR_RUNLOCK(&tracker);
-   KASSERT(counter < NETISR_MAXPROT,
+   KASSERT(counter <= NETISR_MAXPROT,
("sysctl_netisr_proto: counter too big (%d)", counter));
error = SYSCTL_OUT(req, snp_array, sizeof(*snp_array) * counter);
free(snp_array, M_TEMP);
@@ -1212,7 +1212,7 @@ sysctl_netisr_workstream(SYSCTL_HANDLER_
counter++;
}
NETISR_RUNLOCK(&tracker);
-   KASSERT(counter < MAXCPU,
+   KASSERT(counter <= MAXCPU,
("sysctl_netisr_workstream: counter too big (%d)", counter));
error = SYSCTL_OUT(req, snws_array, sizeof(*snws_array) * counter);
free(snws_array, M_TEMP);
@@ -1273,7 +1273,7 @@ sysctl_netisr_work(SYSCTL_HANDLER_ARGS)
}
NWS_UNLOCK(nwsp);
}
-   KASSERT(counter < MAXCPU * NETISR_MAXPROT,
+   KASSERT(counter <= MAXCPU * NETISR_MAXPROT,
("sysctl_netisr_work: counter too big (%d)", counter));
NETISR_RUNLOCK(&tracker);
error = SYSCTL_OUT(req, snw_array, sizeof(*snw_array) * counter);
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217451 - head/sys/powerpc/aim

2011-01-15 Thread Andreas Tobler

Author: andreast
Date: Sat Jan 15 19:16:05 2011
New Revision: 217451
URL: http://svn.freebsd.org/changeset/base/217451

Log:
  Remove unused variables. Spotted by a cppcheck
  (devel/cppcheck, http://sourceforge.net/projects/cppcheck) run.
  
  Approved by: nwhitehorn (mentor)

Modified:
  head/sys/powerpc/aim/slb.c

Modified: head/sys/powerpc/aim/slb.c
==
--- head/sys/powerpc/aim/slb.c  Sat Jan 15 19:05:06 2011(r217450)
+++ head/sys/powerpc/aim/slb.c  Sat Jan 15 19:16:05 2011(r217451)
@@ -200,9 +200,7 @@ make_intermediate(uint64_t esid, struct 
 uint64_t
 kernel_va_to_slbv(vm_offset_t va)
 {
-   uint64_t esid, slbv;
-
-   esid = (uintptr_t)va >> ADDR_SR_SHFT;
+   uint64_t slbv;
 
/* Set kernel VSID to deterministic value */
slbv = (KERNEL_VSID((uintptr_t)va >> ADDR_SR_SHFT)) << SLBV_VSID_SHIFT;
@@ -485,18 +483,11 @@ slb_uma_real_alloc(uma_zone_t zone, int 
static vm_offset_t realmax = 0;
void *va;
vm_page_t m;
-   int pflags;
 
if (realmax == 0)
realmax = platform_real_maxaddr();
 
*flags = UMA_SLAB_PRIV;
-   if ((wait & (M_NOWAIT|M_USE_RESERVE)) == M_NOWAIT)
-   pflags = VM_ALLOC_INTERRUPT | VM_ALLOC_WIRED;
-   else
-   pflags = VM_ALLOC_SYSTEM | VM_ALLOC_WIRED;
-   if (wait & M_ZERO)
-   pflags |= VM_ALLOC_ZERO;
 
for (;;) {
m = vm_phys_alloc_contig(1, 0, realmax, PAGE_SIZE,
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217452 - head/sys/dev/iicbus

2011-01-15 Thread Andreas Tobler

Author: andreast
Date: Sat Jan 15 19:16:56 2011
New Revision: 217452
URL: http://svn.freebsd.org/changeset/base/217452

Log:
  Remove unused variable. Spotted by a cppcheck
  (devel/cppcheck, http://sourceforge.net/projects/cppcheck) run.
  
  Approved by: nwhitehorn (mentor)

Modified:
  head/sys/dev/iicbus/max6690.c

Modified: head/sys/dev/iicbus/max6690.c
==
--- head/sys/dev/iicbus/max6690.c   Sat Jan 15 19:16:05 2011
(r217451)
+++ head/sys/dev/iicbus/max6690.c   Sat Jan 15 19:16:56 2011
(r217452)
@@ -209,7 +209,6 @@ max6690_attach(device_t dev)
 static void
 max6690_start(void *xdev)
 {
-   phandle_t child;
struct max6690_softc *sc;
struct sysctl_oid *oid, *sensroot_oid;
struct sysctl_ctx_list *ctx;
@@ -222,8 +221,6 @@ max6690_start(void *xdev)
 
sc->sc_nsensors = 0;
 
-   child = ofw_bus_get_node(dev);
-
/* Count the actual number of sensors. */
sc->sc_nsensors = max6690_fill_sensor_prop(dev);
 
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217453 - head/sys/vm

2011-01-15 Thread Alan Cox

Author: alc
Date: Sat Jan 15 19:21:28 2011
New Revision: 217453
URL: http://svn.freebsd.org/changeset/base/217453

Log:
  For some time now, the kernel and kmem objects have been ordinary
  OBJT_PHYS objects.  Thus, there is no need for handling them specially
  in vm_fault().  In fact, this special case handling would have led to
  an assertion failure just before the call to pmap_enter().
  
  Reviewed by:  kib@
  MFC after:6 weeks

Modified:
  head/sys/vm/vm_fault.c

Modified: head/sys/vm/vm_fault.c
==
--- head/sys/vm/vm_fault.c  Sat Jan 15 19:16:56 2011(r217452)
+++ head/sys/vm/vm_fault.c  Sat Jan 15 19:21:28 2011(r217453)
@@ -383,11 +383,8 @@ RetryFault:;
 * found the page ).
 */
vm_page_busy(fs.m);
-   if (fs.m->valid != VM_PAGE_BITS_ALL &&
-   fs.m->object != kernel_object && fs.m->object 
!= kmem_object) {
+   if (fs.m->valid != VM_PAGE_BITS_ALL)
goto readrest;
-   }
-
break;
}
 
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217454 - stable/8/lib/libkvm

2011-01-15 Thread Robert Watson

Author: rwatson
Date: Sat Jan 15 19:23:14 2011
New Revision: 217454
URL: http://svn.freebsd.org/changeset/base/217454

Log:
  Merge r204494, r204511 from head to stable/8:
  
A first cut at teaching libkvm how to deal with dynamic per-CPU storage
(DPCPU):
  
A new API, kvm_dpcpu_setcpu(3), selects the active CPU for the purposes
of DPCPU.  Calls to kvm_nlist(3) will automatically translate DPCPU
symbols and return a pointer to the current CPU's version of the data.
Consumers needing to read the same symbol on several CPUs will invoke a
series of setcpu/nlist calls, one per CPU of interest.
  
This addition makes it possible for tools like netstat(1) to query the
values of DPCPU variables during crashdump analysis, and is based on
similar code handling virtualized global variables.
  
Sponsored by:   Juniper Networks, Inc.
  
Not all programs including kvm.h include the necessary headers to use
u_int, so prefer unsigned int.
  
Pointed out by: bz, kib, Mr Tinderbox

Modified:
  stable/8/lib/libkvm/Makefile
  stable/8/lib/libkvm/kvm.c
  stable/8/lib/libkvm/kvm.h
  stable/8/lib/libkvm/kvm_getpcpu.3
  stable/8/lib/libkvm/kvm_pcpu.c
  stable/8/lib/libkvm/kvm_private.h
Directory Properties:
  stable/8/lib/libkvm/   (props changed)

Modified: stable/8/lib/libkvm/Makefile
==
--- stable/8/lib/libkvm/MakefileSat Jan 15 19:21:28 2011
(r217453)
+++ stable/8/lib/libkvm/MakefileSat Jan 15 19:23:14 2011
(r217454)
@@ -21,6 +21,7 @@ MAN=  kvm.3 kvm_getcptime.3 kvm_geterr.3 
kvm_read.3
 
 MLINKS+=kvm_getpcpu.3 kvm_getmaxcpu.3
+MLINKS+=kvm_getpcpu.3 kvm_dpcpu_setcpu.3
 MLINKS+=kvm_getprocs.3 kvm_getargv.3 kvm_getprocs.3 kvm_getenvv.3
 MLINKS+=kvm_open.3 kvm_close.3 kvm_open.3 kvm_openfiles.3
 MLINKS+=kvm_read.3 kvm_write.3

Modified: stable/8/lib/libkvm/kvm.c
==
--- stable/8/lib/libkvm/kvm.c   Sat Jan 15 19:21:28 2011(r217453)
+++ stable/8/lib/libkvm/kvm.c   Sat Jan 15 19:23:14 2011(r217454)
@@ -416,6 +416,8 @@ _kvm_nlist(kvm_t *kd, struct nlist *nl, 
struct kld_sym_lookup lookup;
int error;
char *prefix = "", symname[1024]; /* XXX-BZ symbol name length limit? */
+   int tried_vnet, tried_dpcpu;
+
/*
 * If we can't use the kld symbol lookup, revert to the
 * slow library call.
@@ -429,6 +431,10 @@ _kvm_nlist(kvm_t *kd, struct nlist *nl, 
error = kvm_fdnlist_prefix(kd, nl, error,
VNET_SYMPREFIX, _kvm_vnet_validaddr);
 
+   if (error > 0 && _kvm_dpcpu_initialized(kd, initialize))
+   error = kvm_fdnlist_prefix(kd, nl, error,
+   "pcpu_entry_", _kvm_dpcpu_validaddr);
+
return (error);
}
 
@@ -437,6 +443,8 @@ _kvm_nlist(kvm_t *kd, struct nlist *nl, 
 * and look it up with a kldsym(2) syscall.
 */
nvalid = 0;
+   tried_vnet = 0;
+   tried_dpcpu = 0;
 again:
for (p = nl; p->n_name && p->n_name[0]; ++p) {
if (p->n_type != N_UNDF)
@@ -464,6 +472,10 @@ again:
!strcmp(prefix, VNET_SYMPREFIX))
p->n_value =
_kvm_vnet_validaddr(kd, lookup.symvalue);
+   else if (_kvm_dpcpu_initialized(kd, initialize) &&
+   !strcmp(prefix, "pcpu_entry_"))
+   p->n_value =
+   _kvm_dpcpu_validaddr(kd, lookup.symvalue);
else
p->n_value = lookup.symvalue;
++nvalid;
@@ -473,14 +485,19 @@ again:
 
/*
 * Check the number of entries that weren't found. If they exist,
-* try again with a prefix for virtualized symbol names.
+* try again with a prefix for virtualized or DPCPU symbol names.
 */
error = ((p - nl) - nvalid);
-   if (error && _kvm_vnet_initialized(kd, initialize) &&
-   strcmp(prefix, VNET_SYMPREFIX)) {
+   if (error && _kvm_vnet_initialized(kd, initialize) && !tried_vnet) {
+   tried_vnet = 1;
prefix = VNET_SYMPREFIX;
goto again;
}
+   if (error && _kvm_dpcpu_initialized(kd, initialize) && !tried_dpcpu) {
+   tried_dpcpu = 1;
+   prefix = "pcpu_entry_";
+   goto again;
+   }
 
/*
 * Return the number of entries that weren't found. If they exist,

Modified: stable/8/lib/libkvm/kvm.h
==
--- stable/8/lib/libkvm/kvm.h   Sat Jan 15 19:21:28 2011(r217453)
+++ stable/8/lib/libkvm/kvm.h   Sat Jan 15 19:23:14 2011(r217454)
@@

svn commit: r217455 - stable/8/sys/net

2011-01-15 Thread Robert Watson

Author: rwatson
Date: Sat Jan 15 19:27:22 2011
New Revision: 217455
URL: http://svn.freebsd.org/changeset/base/217455

Log:
  Merge r203913 from head to stable/8 (original change by pjd):
  
Mark various sysctls also as tunables.
  
Reviewed by:rwatson

Modified:
  stable/8/sys/net/netisr.c
Directory Properties:
  stable/8/sys/   (props changed)
  stable/8/sys/amd64/include/xen/   (props changed)
  stable/8/sys/cddl/contrib/opensolaris/   (props changed)
  stable/8/sys/contrib/dev/acpica/   (props changed)
  stable/8/sys/contrib/pf/   (props changed)

Modified: stable/8/sys/net/netisr.c
==
--- stable/8/sys/net/netisr.c   Sat Jan 15 19:23:14 2011(r217454)
+++ stable/8/sys/net/netisr.c   Sat Jan 15 19:27:22 2011(r217455)
@@ -160,13 +160,13 @@ SYSCTL_INT(_net_isr, OID_AUTO, direct, C
  */
 static int netisr_maxthreads = -1; /* Max number of threads. */
 TUNABLE_INT("net.isr.maxthreads", &netisr_maxthreads);
-SYSCTL_INT(_net_isr, OID_AUTO, maxthreads, CTLFLAG_RD,
+SYSCTL_INT(_net_isr, OID_AUTO, maxthreads, CTLFLAG_RDTUN,
 &netisr_maxthreads, 0,
 "Use at most this many CPUs for netisr processing");
 
 static int netisr_bindthreads = 0; /* Bind threads to CPUs. */
 TUNABLE_INT("net.isr.bindthreads", &netisr_bindthreads);
-SYSCTL_INT(_net_isr, OID_AUTO, bindthreads, CTLFLAG_RD,
+SYSCTL_INT(_net_isr, OID_AUTO, bindthreads, CTLFLAG_RDTUN,
 &netisr_bindthreads, 0, "Bind netisr threads to CPUs.");
 
 /*
@@ -176,7 +176,7 @@ SYSCTL_INT(_net_isr, OID_AUTO, bindthrea
 #defineNETISR_DEFAULT_MAXQLIMIT10240
 static u_int   netisr_maxqlimit = NETISR_DEFAULT_MAXQLIMIT;
 TUNABLE_INT("net.isr.maxqlimit", &netisr_maxqlimit);
-SYSCTL_INT(_net_isr, OID_AUTO, maxqlimit, CTLFLAG_RD,
+SYSCTL_INT(_net_isr, OID_AUTO, maxqlimit, CTLFLAG_RDTUN,
 &netisr_maxqlimit, 0,
 "Maximum netisr per-protocol, per-CPU queue depth.");
 
@@ -188,7 +188,7 @@ SYSCTL_INT(_net_isr, OID_AUTO, maxqlimit
 #defineNETISR_DEFAULT_DEFAULTQLIMIT256
 static u_int   netisr_defaultqlimit = NETISR_DEFAULT_DEFAULTQLIMIT;
 TUNABLE_INT("net.isr.defaultqlimit", &netisr_defaultqlimit);
-SYSCTL_INT(_net_isr, OID_AUTO, defaultqlimit, CTLFLAG_RD,
+SYSCTL_INT(_net_isr, OID_AUTO, defaultqlimit, CTLFLAG_RDTUN,
 &netisr_defaultqlimit, 0,
 "Default netisr per-protocol, per-CPU queue limit if not set by protocol");
 
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217456 - stable/8/sys/net

2011-01-15 Thread Robert Watson

Author: rwatson
Date: Sat Jan 15 19:36:47 2011
New Revision: 217456
URL: http://svn.freebsd.org/changeset/base/217456

Log:
  Merge from head to stable/8:
  
Refine netisr.c comments a bit.

Modified:
  stable/8/sys/net/netisr.c
Directory Properties:
  stable/8/sys/   (props changed)
  stable/8/sys/amd64/include/xen/   (props changed)
  stable/8/sys/cddl/contrib/opensolaris/   (props changed)
  stable/8/sys/contrib/dev/acpica/   (props changed)
  stable/8/sys/contrib/pf/   (props changed)

Modified: stable/8/sys/net/netisr.c
==
--- stable/8/sys/net/netisr.c   Sat Jan 15 19:27:22 2011(r217455)
+++ stable/8/sys/net/netisr.c   Sat Jan 15 19:36:47 2011(r217456)
@@ -36,13 +36,13 @@ __FBSDID("$FreeBSD$");
  * dispatched) and asynchronous (deferred dispatch) processing of packets by
  * registered protocol handlers.  Callers pass a protocol identifier and
  * packet to netisr, along with a direct dispatch hint, and work will either
- * be immediately processed with the registered handler, or passed to a
- * kernel software interrupt (SWI) thread for deferred dispatch.  Callers
- * will generally select one or the other based on:
+ * be immediately processed by the registered handler, or passed to a
+ * software interrupt (SWI) thread for deferred dispatch.  Callers will
+ * generally select one or the other based on:
  *
- * - Might directly dispatching a netisr handler lead to code reentrance or
+ * - Whether directly dispatching a netisr handler lead to code reentrance or
  *   lock recursion, such as entering the socket code from the socket code.
- * - Might directly dispatching a netisr handler lead to recursive
+ * - Whether directly dispatching a netisr handler lead to recursive
  *   processing, such as when decapsulating several wrapped layers of tunnel
  *   information (IPSEC within IPSEC within ...).
  *
@@ -58,9 +58,9 @@ __FBSDID("$FreeBSD$");
  * more than one flow.
  *
  * netisr supports several policy variations, represented by the
- * NETISR_POLICY_* constants, allowing protocols to play a varying role in
+ * NETISR_POLICY_* constants, allowing protocols to play various roles in
  * identifying flows, assigning work to CPUs, etc.  These are described in
- * detail in netisr.h.
+ * netisr.h.
  */
 
 #include "opt_ddb.h"
@@ -105,7 +105,7 @@ __FBSDID("$FreeBSD$");
  *
  * Note: the NETISR_LOCKING define controls whether read locks are acquired
  * in packet processing paths requiring netisr registration stability.  This
- * is disabled by default as it can lead to a measurable performance
+ * is disabled by default as it can lead to measurable performance
  * degradation even with rmlocks (3%-6% for loopback ping-pong traffic), and
  * because netisr registration and unregistration is extremely rare at
  * runtime.  If it becomes more common, this decision should be revisited.
@@ -170,8 +170,9 @@ SYSCTL_INT(_net_isr, OID_AUTO, bindthrea
 &netisr_bindthreads, 0, "Bind netisr threads to CPUs.");
 
 /*
- * Limit per-workstream queues to at most net.isr.maxqlimit, both for initial
- * configuration and later modification using netisr_setqlimit().
+ * Limit per-workstream mbuf queue limits s to at most net.isr.maxqlimit,
+ * both for initial configuration and later modification using
+ * netisr_setqlimit().
  */
 #defineNETISR_DEFAULT_MAXQLIMIT10240
 static u_int   netisr_maxqlimit = NETISR_DEFAULT_MAXQLIMIT;
@@ -181,9 +182,9 @@ SYSCTL_INT(_net_isr, OID_AUTO, maxqlimit
 "Maximum netisr per-protocol, per-CPU queue depth.");
 
 /*
- * The default per-workstream queue limit for protocols that don't initialize
- * the nh_qlimit field of their struct netisr_handler.  If this is set above
- * netisr_maxqlimit, we truncate it to the maximum during boot.
+ * The default per-workstream mbuf queue limit for protocols that don't
+ * initialize the nh_qlimit field of their struct netisr_handler.  If this is
+ * set above netisr_maxqlimit, we truncate it to the maximum during boot.
  */
 #defineNETISR_DEFAULT_DEFAULTQLIMIT256
 static u_int   netisr_defaultqlimit = NETISR_DEFAULT_DEFAULTQLIMIT;
@@ -241,12 +242,14 @@ struct netisr_work {
 };
 
 /*
- * Workstreams hold a set of ordered work across each protocol, and are
+ * Workstreams hold a queue of ordered work across each protocol, and are
  * described by netisr_workstream.  Each workstream is associated with a
  * worker thread, which in turn is pinned to a CPU.  Work associated with a
  * workstream can be processd in other threads during direct dispatch;
  * concurrent processing is prevented by the NWS_RUNNING flag, which
- * indicates that a thread is already processing the work queue.
+ * indicates that a thread is already processing the work queue.  It is
+ * important to prevent a directly dispatched packet from "skipping ahead" of
+ * work already in the workstream queue.
  */
 struct netisr_workstream {
struct intr_ev

svn commit: r217457 - stable/8/sys/net

2011-01-15 Thread Robert Watson

Author: rwatson
Date: Sat Jan 15 19:46:36 2011
New Revision: 217457
URL: http://svn.freebsd.org/changeset/base/217457

Log:
  Merge r204497, r204499 from head to stable/8:
  
Changes to support crashdump analysis of netisr:
  
- Rename the netisr protocol registration array, 'np' to 'netisr_proto',
  in order to reduce the chances of symbol name collisions.  It remains
  statically defined, but it will be looked up by netstat(1).
  
- Move certain internal structure definitions from netisr.c to
  netisr_internal.h so that netstat(1) can find them.  They remain
  private, and should not be used for any other purpose (for example,
  they should not be used by kernel modules, which must instead use the
  public interfaces in netisr.h).
  
- Store a kernel-compiled version of NETISR_MAXPROT in the global variable
  netisr_maxprot, and export via a sysctl, so that it is available for use
  by netstat(1).  This is especially important for crashdump
  interpretation, where the size of the workstream structure is determined
  by the maximum number of protocols compiled into the kernel.
  
Sponsored by:   Juniper Networks
  
Whitespace tweak.

Added:
  stable/8/sys/net/netisr_internal.h
 - copied unchanged from r204497, head/sys/net/netisr_internal.h
Modified:
  stable/8/sys/net/netisr.c
Directory Properties:
  stable/8/sys/   (props changed)
  stable/8/sys/amd64/include/xen/   (props changed)
  stable/8/sys/cddl/contrib/opensolaris/   (props changed)
  stable/8/sys/contrib/dev/acpica/   (props changed)
  stable/8/sys/contrib/pf/   (props changed)

Modified: stable/8/sys/net/netisr.c
==
--- stable/8/sys/net/netisr.c   Sat Jan 15 19:36:47 2011(r217456)
+++ stable/8/sys/net/netisr.c   Sat Jan 15 19:46:36 2011(r217457)
@@ -87,9 +87,11 @@ __FBSDID("$FreeBSD$");
 #include 
 #endif
 
+#define_WANT_NETISR_INTERNAL   /* Enable definitions from 
netisr_internal.h */
 #include 
 #include 
 #include 
+#include 
 #include 
 
 /*-
@@ -99,7 +101,7 @@ __FBSDID("$FreeBSD$");
  *
  * The following data structures and fields are protected by this lock:
  *
- * - The np array, including all fields of struct netisr_proto.
+ * - The netisr_proto array, including all fields of struct netisr_proto.
  * - The nws array, including all fields of struct netisr_worker.
  * - The nws_array array.
  *
@@ -194,79 +196,23 @@ SYSCTL_INT(_net_isr, OID_AUTO, defaultql
 "Default netisr per-protocol, per-CPU queue limit if not set by protocol");
 
 /*
- * Each protocol is described by a struct netisr_proto, which holds all
- * global per-protocol information.  This data structure is set up by
- * netisr_register(), and derived from the public struct netisr_handler.
+ * Store and export the compile-time constant NETISR_MAXPROT limit on the
+ * number of protocols that can register with netisr at a time.  This is
+ * required for crashdump analysis, as it sizes netisr_proto[].
  */
-struct netisr_proto {
-   const char  *np_name;   /* Character string protocol name. */
-   netisr_handler_t *np_handler;   /* Protocol handler. */
-   netisr_m2flow_t *np_m2flow; /* Query flow for untagged packet. */
-   netisr_m2cpuid_t *np_m2cpuid;   /* Query CPU to process packet on. */
-   netisr_drainedcpu_t *np_drainedcpu; /* Callback when drained a queue. */
-   u_intnp_qlimit; /* Maximum per-CPU queue depth. */
-   u_intnp_policy; /* Work placement policy. */
-};
-
-#defineNETISR_MAXPROT  16  /* Compile-time limit. 
*/
+static u_int   netisr_maxprot = NETISR_MAXPROT;
+SYSCTL_INT(_net_isr, OID_AUTO, maxprot, CTLFLAG_RD,
+&netisr_maxprot, 0,
+"Compile-time limit on the number of protocols supported by netisr.");
 
 /*
- * The np array describes all registered protocols, indexed by protocol
- * number.
+ * The netisr_proto array describes all registered protocols, indexed by
+ * protocol number.  See netisr_internal.h for more details.
  */
-static struct netisr_proto np[NETISR_MAXPROT];
+static struct netisr_proto netisr_proto[NETISR_MAXPROT];
 
 /*
- * Protocol-specific work for each workstream is described by struct
- * netisr_work.  Each work descriptor consists of an mbuf queue and
- * statistics.
- */
-struct netisr_work {
-   /*
-* Packet queue, linked by m_nextpkt.
-*/
-   struct mbuf *nw_head;
-   struct mbuf *nw_tail;
-   u_intnw_len;
-   u_intnw_qlimit;
-   u_intnw_watermark;
-
-   /*
-* Statistics -- written unlocked, but mostly from curcpu.
-*/
-   u_int64_tnw_dispatched; /* Number of direct dispatches. */
-   u_int64_tnw_hybrid_dispatched; /* "" hybrid dispatches. */
-   u_int64_tnw_qdrops; /* "" drops. */
-   u_int64_t

svn commit: r217458 - stable/8/usr.bin/netstat

2011-01-15 Thread Robert Watson

Author: rwatson
Date: Sat Jan 15 19:49:08 2011
New Revision: 217458
URL: http://svn.freebsd.org/changeset/base/217458

Log:
  Merge r204499, r204515 from head to stable/8:
  
Teach netstat -Q to work with -N and -M by adding libkvm versions of data
query routines.  This code is necessarily more fragile in the presence of
kernel changes than querying the kernel via sysctl (the default), but
useful when investigating crashes or live kernel state via firewire.
  
Sponsored by:   Juniper Networks
  
Prefer vocabulary of 'Current' and 'Limit' to 'Value' and 'Maximum' in
netstat -Q.
  
Sponsored by:   Juniper Networks

Modified:
  stable/8/usr.bin/netstat/main.c
  stable/8/usr.bin/netstat/netisr.c
  stable/8/usr.bin/netstat/netstat.h
Directory Properties:
  stable/8/usr.bin/netstat/   (props changed)

Modified: stable/8/usr.bin/netstat/main.c
==
--- stable/8/usr.bin/netstat/main.c Sat Jan 15 19:46:36 2011
(r217457)
+++ stable/8/usr.bin/netstat/main.c Sat Jan 15 19:49:08 2011
(r217458)
@@ -539,9 +539,11 @@ main(int argc, char *argv[])
exit(0);
}
if (Qflag) {
-   if (!live)
-   usage();
-   netisr_stats();
+   if (!live) {
+   if (kread(0, NULL, 0) == 0)
+   netisr_stats(kvmd);
+   } else
+   netisr_stats(NULL);
exit(0);
}
 #if 0

Modified: stable/8/usr.bin/netstat/netisr.c
==
--- stable/8/usr.bin/netstat/netisr.c   Sat Jan 15 19:46:36 2011
(r217457)
+++ stable/8/usr.bin/netstat/netisr.c   Sat Jan 15 19:49:08 2011
(r217458)
@@ -31,15 +31,22 @@
 
 __FBSDID("$FreeBSD$");
 
-#include 
+#include 
 #include 
 
+#include 
+#include 
+
+#define_WANT_NETISR_INTERNAL
 #include 
+#include 
 
 #include 
+#include 
 #include 
 #include 
 #include 
+#include 
 
 #include "netstat.h"
 
@@ -65,6 +72,90 @@ static u_int  workstream_array_len;
 static struct sysctl_netisr_work   *work_array;
 static u_intwork_array_len;
 
+static u_int   *nws_array;
+
+static u_intmaxprot;
+
+static void
+netisr_load_kvm_uint(kvm_t *kd, char *name, u_int *p)
+{
+   struct nlist nl[] = {
+   { .n_name = name },
+   { .n_name = NULL },
+   };
+   int ret;
+
+   ret = kvm_nlist(kd, nl);
+   if (ret < 0)
+   errx(-1, "%s: kvm_nlist(%s): %s", __func__, name,
+   kvm_geterr(kd));
+   if (ret != 0)
+   errx(-1, "%s: kvm_nlist(%s): unresolved symbol", __func__,
+   name);
+   if (kvm_read(kd, nl[0].n_value, p, sizeof(*p)) != sizeof(*p))
+   errx(-1, "%s: kvm_read(%s): %s", __func__, name,
+   kvm_geterr(kd));
+}
+
+/*
+ * Load a nul-terminated string from KVM up to 'limit', guarantee that the
+ * string in local memory is nul-terminated.
+ */
+static void
+netisr_load_kvm_string(kvm_t *kd, uintptr_t addr, char *dest, u_int limit)
+{
+   u_int i;
+
+   for (i = 0; i < limit; i++) {
+   if (kvm_read(kd, addr + i, &dest[i], sizeof(dest[i])) !=
+   sizeof(dest[i]))
+   err(-1, "%s: kvm_read: %s", __func__,
+   kvm_geterr(kd));
+   if (dest[i] == '\0')
+   break;
+   }
+   dest[limit - 1] = '\0';
+}
+
+static const char *
+netisr_proto2name(u_int proto)
+{
+   u_int i;
+
+   for (i = 0; i < proto_array_len; i++) {
+   if (proto_array[i].snp_proto == proto)
+   return (proto_array[i].snp_name);
+   }
+   return ("unknown");
+}
+
+static int
+netisr_protoispresent(u_int proto)
+{
+   u_int i;
+
+   for (i = 0; i < proto_array_len; i++) {
+   if (proto_array[i].snp_proto == proto)
+   return (1);
+   }
+   return (0);
+}
+
+static void
+netisr_load_kvm_config(kvm_t *kd)
+{
+
+   netisr_load_kvm_uint(kd, "_netisr_bindthreads", &bindthreads);
+   netisr_load_kvm_uint(kd, "_netisr_maxthreads", &maxthreads);
+   netisr_load_kvm_uint(kd, "_nws_count", &numthreads);
+
+   netisr_load_kvm_uint(kd, "_netisr_defaultqlimit", &defaultqlimit);
+   netisr_load_kvm_uint(kd, "_netisr_maxqlimit", &maxqlimit);
+
+   netisr_load_kvm_uint(kd, "_netisr_direct", &direct);
+   netisr_load_kvm_uint(kd, "_netisr_direct_force", &direct_force);
+}
+
 static void
 netisr_load_sysctl_uint(const char *name, u_int *p)
 {
@@ -78,7 +169,7 @@ netisr_load_sysctl_uint(const char *name
 }
 
 static void
-netisr_load_config(void)
+netisr_load_sysctl_config(void)
 {
 
netisr_load_sysctl_uint("net.is

svn commit: r217459 - head/sys/powerpc/include

2011-01-15 Thread Marcel Moolenaar

Author: marcel
Date: Sat Jan 15 20:18:08 2011
New Revision: 217459
URL: http://svn.freebsd.org/changeset/base/217459

Log:
  Don't redefine MODINFOMD_BOOTINFO as MODINFOMD_DTBP. This
  breaks support for older loaders. Add MODINFOMD_DTBP as
  a new tag instead.

Modified:
  head/sys/powerpc/include/metadata.h

Modified: head/sys/powerpc/include/metadata.h
==
--- head/sys/powerpc/include/metadata.h Sat Jan 15 19:49:08 2011
(r217458)
+++ head/sys/powerpc/include/metadata.h Sat Jan 15 20:18:08 2011
(r217459)
@@ -32,6 +32,7 @@
 #defineMODINFOMD_ENVP  0x1001
 #defineMODINFOMD_HOWTO 0x1002
 #defineMODINFOMD_KERNEND   0x1003
-#defineMODINFOMD_DTBP  0x1004
+#defineMODINFOMD_BOOTINFO  0x1004
+#defineMODINFOMD_DTBP  0x1005
 
 #endif /* !_MACHINE_METADATA_H_ */
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217461 - in head: bin/sh tools/regression/bin/sh/builtins

2011-01-15 Thread Jilles Tjoelker

Author: jilles
Date: Sat Jan 15 21:09:00 2011
New Revision: 217461
URL: http://svn.freebsd.org/changeset/base/217461

Log:
  sh: Fix some things about -- in trap:
  * Make 'trap --' do the same as 'trap' instead of nothing.
  * Make '--' stop option processing (note that '-' action is not an option).
  
  Side effect: The error message for an unknown option is different.

Added:
  head/tools/regression/bin/sh/builtins/trap5.0   (contents, props changed)
Modified:
  head/bin/sh/trap.c

Modified: head/bin/sh/trap.c
==
--- head/bin/sh/trap.c  Sat Jan 15 20:37:56 2011(r217460)
+++ head/bin/sh/trap.c  Sat Jan 15 21:09:00 2011(r217461)
@@ -153,8 +153,18 @@ trapcmd(int argc, char **argv)
char *action;
int signo;
int errors = 0;
+   int i;
+
+   while ((i = nextopt("l")) != '\0') {
+   switch (i) {
+   case 'l':
+   printsignals();
+   return (0);
+   }
+   }
+   argv = argptr;
 
-   if (argc <= 1) {
+   if (*argv == NULL) {
for (signo = 0 ; signo < sys_nsig ; signo++) {
if (signo < NSIG && trap[signo] != NULL) {
out1str("trap -- ");
@@ -171,19 +181,12 @@ trapcmd(int argc, char **argv)
return 0;
}
action = NULL;
-   if (*++argv && strcmp(*argv, "--") == 0)
-   argv++;
if (*argv && sigstring_to_signum(*argv) == -1) {
-   if ((*argv)[0] != '-') {
-   action = *argv;
+   if (strcmp(*argv, "-") == 0)
argv++;
-   } else if ((*argv)[1] == '\0') {
+   else {
+   action = *argv;
argv++;
-   } else if ((*argv)[1] == 'l' && (*argv)[2] == '\0') {
-   printsignals();
-   return 0;
-   } else {
-   error("bad option %s", *argv);
}
}
while (*argv) {

Added: head/tools/regression/bin/sh/builtins/trap5.0
==
--- /dev/null   00:00:00 1970   (empty, because file is newly added)
+++ head/tools/regression/bin/sh/builtins/trap5.0   Sat Jan 15 21:09:00 
2011(r217461)
@@ -0,0 +1,19 @@
+# $FreeBSD$
+
+set -e
+trap - USR1
+initial=$(trap)
+trap -- -l USR1
+added=$(trap)
+[ -n "$added" ]
+trap - USR1
+second=$(trap)
+[ "$initial" = "$second" ]
+eval "$added"
+added2=$(trap)
+added3=$(trap --)
+[ "$added" = "$added2" ]
+[ "$added2" = "$added3" ]
+trap -- - USR1
+third=$(trap)
+[ "$initial" = "$third" ]
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217463 - head/sys/vm

2011-01-15 Thread Konstantin Belousov

Author: kib
Date: Sat Jan 15 21:56:38 2011
New Revision: 217463
URL: http://svn.freebsd.org/changeset/base/217463

Log:
  For consistency, use kernel_object instead of &kernel_object_store
  when initializing the object mutex. Do the same for kmem_object.
  
  Discussed with:   alc
  MFC after:1 week

Modified:
  head/sys/vm/vm_object.c

Modified: head/sys/vm/vm_object.c
==
--- head/sys/vm/vm_object.c Sat Jan 15 21:24:39 2011(r217462)
+++ head/sys/vm/vm_object.c Sat Jan 15 21:56:38 2011(r217463)
@@ -244,7 +244,7 @@ vm_object_init(void)
TAILQ_INIT(&vm_object_list);
mtx_init(&vm_object_list_mtx, "vm object_list", NULL, MTX_DEF);

-   VM_OBJECT_LOCK_INIT(&kernel_object_store, "kernel object");
+   VM_OBJECT_LOCK_INIT(kernel_object, "kernel object");
_vm_object_allocate(OBJT_PHYS, OFF_TO_IDX(VM_MAX_KERNEL_ADDRESS - 
VM_MIN_KERNEL_ADDRESS),
kernel_object);
 #if VM_NRESERVLEVEL > 0
@@ -252,7 +252,7 @@ vm_object_init(void)
kernel_object->pg_color = (u_short)atop(VM_MIN_KERNEL_ADDRESS);
 #endif
 
-   VM_OBJECT_LOCK_INIT(&kmem_object_store, "kmem object");
+   VM_OBJECT_LOCK_INIT(kmem_object, "kmem object");
_vm_object_allocate(OBJT_PHYS, OFF_TO_IDX(VM_MAX_KERNEL_ADDRESS - 
VM_MIN_KERNEL_ADDRESS),
kmem_object);
 #if VM_NRESERVLEVEL > 0
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217464 - head/share/man/man4

2011-01-15 Thread Marius Strobl

Author: marius
Date: Sat Jan 15 22:07:08 2011
New Revision: 217464
URL: http://svn.freebsd.org/changeset/base/217464

Log:
  Add a manual page for rgephy(4) and reference it as appropriate. The
  motivation for having rgephy.4 is to document the special media option
  add in r217415.
  
  MFC after:3 days

Added:
  head/share/man/man4/rgephy.4   (contents, props changed)
Modified:
  head/share/man/man4/Makefile
  head/share/man/man4/miibus.4
  head/share/man/man4/nfe.4
  head/share/man/man4/re.4

Modified: head/share/man/man4/Makefile
==
--- head/share/man/man4/MakefileSat Jan 15 21:56:38 2011
(r217463)
+++ head/share/man/man4/MakefileSat Jan 15 22:07:08 2011
(r217464)
@@ -337,6 +337,7 @@ MAN=aac.4 \
random.4 \
rc.4 \
re.4 \
+   rgephy.4 \
rl.4 \
rndtest.4 \
route.4 \

Modified: head/share/man/man4/miibus.4
==
--- head/share/man/man4/miibus.4Sat Jan 15 21:56:38 2011
(r217463)
+++ head/share/man/man4/miibus.4Sat Jan 15 22:07:08 2011
(r217464)
@@ -8,7 +8,7 @@
 .\"
 .\" $FreeBSD$
 .\"
-.Dd December 30, 2010
+.Dd January 15, 2011
 .Dt MIIBUS 4
 .Os
 .Sh NAME
@@ -162,6 +162,7 @@ but as a result are not well behaved new
 .Xr nve 4 ,
 .Xr pcn 4 ,
 .Xr re 4 ,
+.Xr rgephy 4 ,
 .Xr rl 4 ,
 .Xr rue 4 ,
 .Xr sf 4 ,

Modified: head/share/man/man4/nfe.4
==
--- head/share/man/man4/nfe.4   Sat Jan 15 21:56:38 2011(r217463)
+++ head/share/man/man4/nfe.4   Sat Jan 15 22:07:08 2011(r217464)
@@ -16,7 +16,7 @@
 .\"
 .\" $FreeBSD$
 .\"
-.Dd October 2, 2008
+.Dd January, 2011
 .Dt NFE 4
 .Os
 .Sh NAME
@@ -174,6 +174,7 @@ before a change takes effect.
 .Xr netintro 4 ,
 .Xr pci 4 ,
 .Xr polling 4 ,
+.Xr rgephy 4 ,
 .Xr sysctl 8 ,
 .Xr ifconfig 8
 .Sh HISTORY

Modified: head/share/man/man4/re.4
==
--- head/share/man/man4/re.4Sat Jan 15 21:56:38 2011(r217463)
+++ head/share/man/man4/re.4Sat Jan 15 22:07:08 2011(r217464)
@@ -30,7 +30,7 @@
 .\"
 .\" $FreeBSD$
 .\"
-.Dd December 2, 2010
+.Dd January, 2011
 .Dt RE 4
 .Os
 .Sh NAME
@@ -213,6 +213,7 @@ the network connection (cable).
 .Xr netintro 4 ,
 .Xr ng_ether 4 ,
 .Xr polling 4 ,
+.Xr rgephy 4 ,
 .Xr vlan 4 ,
 .Xr ifconfig 8
 .Rs

Added: head/share/man/man4/rgephy.4
==
--- /dev/null   00:00:00 1970   (empty, because file is newly added)
+++ head/share/man/man4/rgephy.4Sat Jan 15 22:07:08 2011
(r217464)
@@ -0,0 +1,96 @@
+.\"
+.\" Copyright (c) 2011 Marius Strobl 
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions
+.\" are met:
+.\" 1. Redistributions of source code must retain the above copyright
+.\"notice, this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright
+.\"notice, this list of conditions and the following disclaimer in the
+.\"documentation and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
+.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+.\" ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
+.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+.\" SUCH DAMAGE.
+.\"
+.\" $FreeBSD$
+.\"
+.Dd January 15 2011 $
+.Dt RGEPHY 4
+.Os
+.Sh NAME
+.Nm rgephy
+.Nd RealTek RTL8168/8169/8110/8211 series 10/100/1000 Gigabit Ethernet PHY 
driver
+.Sh SYNOPSIS
+To compile all available PHY drivers into the kernel,
+place the following line in your kernel configuration file:
+.Bd -ragged -offset indent
+.Cd "device miibus"
+.Ed
+.Pp
+Alternatively, to selectively compile this driver into the kernel,
+place the following lines in your kernel configuration file instead:
+.Bd -ragged -offset indent
+.Cd "device mii"
+.Cd "device rgephy"
+.Ed
+.Sh DESCRIPTION
+The
+.Nm
+driver supports the RealTek RTL8168, RTL8169, RTL8110 and RTL8211 series
+integrated 10/100/1000 Gigabit Ethernet PHYs.
+.Pp
+In order to get a list of

Re: svn commit: r217369 - in head/sys: cam/scsi sys

2011-01-15 Thread Bruce Evans


On Sat, 15 Jan 2011 m...@freebsd.org wrote:


On Sat, Jan 15, 2011 at 6:55 AM, Bruce Evans  wrote:



SYSCTL_I() works even better that I first thought. ?It automatically
gives support for all typedefed integral types. ?No SYSCTL_FOO_T()s
...


Grrr, my sentence breaks of 2 spaces are being echoed as 1 space and 1
hard \xa0.


The printing is done entirely in user-space, so it's not too bad.  I
had figured to upcast everything to [u]intmax_t anyways, and what to
cast from would just be [u]int32_t and [u]int64_t depending on
reported size.  Anything smaller should be copyout(9)'d as a 4 bit
quantity.


I think it shouldn't be converted in the kernel.  Applications using
sysctl already have to deal with fields shorter than int in structs
returned by sysctl().  They don't need to deal with integers shorter
than int only since sysctl() doesn't support such integers yet.  Short
integers could still be returned as essentially themselves by packing
them into a struct with nothing else.


But, there's two factors at play here.  The first is sysctl(8) which
asks for the size when reading and manages it.  The second is
sysctl(2) where we have code like SCTL_MASK32 so a long on a 64-bit
kernel looks like a long to a 32-bit app.  It would be simpler to
expose this as an 8-byte quantity on 64-bit kernel, and sysctl(8) will
deal with this just fine, but that would break some uses of sysctl(2).


What are the uses?  I guess they are not-so-hard-coding data types as
int and long, etc., and hard-coding them as int32_t and int64_t.  Then
you get an error when the kernel returns an unexpected length, but
unexpected lengths are expected for 32-bit apps on 64-bit kernels.

BTW, short writes are still horribly broken.  Suppose there is a size
mismatch causing an application tries to write 32 bits to a 64-bit
kernel variable.  Then no error is detected, and the kernel variable
ois left with garbage in its top bits.  The error detection last worked
with 4.4BSD sysctl, and is still documented to work:

%  [EINVAL]   A non-null newp is given and its specified length in
% newlen is too large or too small.

I think the case where newlen is too large still works.

I guess the bug is potentially larger for 32 bit compat applications,
but it is rarely seen for those too since only system management
applications should be writing to kernel variables and there is no
reason to run such applications in 32 bit mode.  The bug affects mainly
cases where there is a kernel type mismatch, like using sysctl_handle_int()
on a 64-bit variable.  I forget how short reads were not detected as
errors.  It still takes the size in the sysctl data to be for 64 bits
for this error to be detectable for either read or write.  But your
recent changes should fix all such type mismatches.


However, I *think* it is safe to check the sysctl_req's oldidx/oldlen
to see what size the user-space expects and cast a kernel long to int
before SYSCTL_OUT.


Maybe if the value fits.  The application might be probing for a size
that works.  Then it shouldn't care what size the value is returned in,
provided the value fits.  Writing is only slightly more delicate.  The
application must use a size in which the value fits.  Then the kernel
can expand the size if necessary.  It just has to be careful to fill
the top bits and not have sign extension bugs.  sysctl_handle_i() can
probably handle both directions.  Lower-level sysctl code should still
return an error if there is a size mismatch.

If a value doesn't fit, then the application will just have to detect
the error and try a larger size (or fail if it doesn't support larger
size).  An example might be a 32-bit app reading 64-bit kernel pointers.
These probably have the high bits set, so they won't fit in 32-bit sizes.
But the application can easily read them using 64-bit sizes.  Integers
and pointers larger than 64 bits would be more interesting, since a
compat application might not have any integer data type larger enough
to represent them.

So I changed my mind about converting in the kernel.  Just try to convert
to whatever size the application is trying to use.  Don't even force >= 32
bit sizes on applications.


$WORK has gotten busy so I haven't had a chance to work on this much
but I'm hopeful that the missus will watch the kids this weekend and
allow me to hack.


You seem to have found time :-).

Bruce___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217468 - head/share/man/man4

2011-01-15 Thread Marius Strobl

Author: marius
Date: Sun Jan 16 04:14:56 2011
New Revision: 217468
URL: http://svn.freebsd.org/changeset/base/217468

Log:
  Reference rgephy.4 in man pages of additional MAC drivers also known to
  come in combination with these PHYs.
  
  Submitted by: yongari
  MFC after:3 days

Modified:
  head/share/man/man4/axe.4
  head/share/man/man4/nve.4
  head/share/man/man4/sge.4

Modified: head/share/man/man4/axe.4
==
--- head/share/man/man4/axe.4   Sun Jan 16 00:45:06 2011(r217467)
+++ head/share/man/man4/axe.4   Sun Jan 16 04:14:56 2011(r217468)
@@ -30,7 +30,7 @@
 .\"
 .\" $FreeBSD$
 .\"
-.Dd January 20, 2008
+.Dd January 16, 2011
 .Dt AXE 4
 .Os
 .Sh NAME
@@ -206,6 +206,7 @@ The driver failed to allocate an mbuf fo
 .Xr miibus 4 ,
 .Xr netintro 4 ,
 .Xr ng_ether 4 ,
+.Xr rgephy 4 ,
 .Xr ifconfig 8
 .Rs
 .%T "ASIX AX88172 AX88178 and AX88772 data sheets"

Modified: head/share/man/man4/nve.4
==
--- head/share/man/man4/nve.4   Sun Jan 16 00:45:06 2011(r217467)
+++ head/share/man/man4/nve.4   Sun Jan 16 04:14:56 2011(r217468)
@@ -26,7 +26,7 @@
 .\"
 .\" $FreeBSD$
 .\"
-.Dd December 8, 2007
+.Dd January 16, 2011
 .Dt NVE 4
 .Os
 .Sh NAME
@@ -120,6 +120,7 @@ the network connection (cable).
 .Xr miibus 4 ,
 .Xr netintro 4 ,
 .Xr ng_ether 4 ,
+.Xr rgephy 4 ,
 .Xr ifconfig 8
 .Sh HISTORY
 The

Modified: head/share/man/man4/sge.4
==
--- head/share/man/man4/sge.4   Sun Jan 16 00:45:06 2011(r217467)
+++ head/share/man/man4/sge.4   Sun Jan 16 04:14:56 2011(r217468)
@@ -24,7 +24,7 @@
 .\"
 .\" $FreeBSD$
 .\"
-.Dd May 10, 2010
+.Dd January 16, 2011
 .Dt SGE 4
 .Os
 .Sh NAME
@@ -105,6 +105,7 @@ SiS191 Fast/Gigabit Ethernet controller
 .Xr miibus 4 ,
 .Xr netintro 4 ,
 .Xr ng_ether 4 ,
+.Xr rgephy 4 ,
 .Xr vlan 4 ,
 .Xr ifconfig 8
 .Sh HISTORY
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

Re: svn commit: r217369 - in head/sys: cam/scsi sys

2011-01-15 Thread mdf

On Sat, Jan 15, 2011 at 7:06 PM, Bruce Evans  wrote:
> On Sat, 15 Jan 2011 m...@freebsd.org wrote:
>> On Sat, Jan 15, 2011 at 6:55 AM, Bruce Evans  wrote:
>
>> The printing is done entirely in user-space, so it's not too bad.  I
>> had figured to upcast everything to [u]intmax_t anyways, and what to
>> cast from would just be [u]int32_t and [u]int64_t depending on
>> reported size.  Anything smaller should be copyout(9)'d as a 4 bit
>> quantity.
>
> I think it shouldn't be converted in the kernel.  Applications using
> sysctl already have to deal with fields shorter than int in structs
> returned by sysctl().  They don't need to deal with integers shorter
> than int only since sysctl() doesn't support such integers yet.  Short
> integers could still be returned as essentially themselves by packing
> them into a struct with nothing else.
>
>> But, there's two factors at play here.  The first is sysctl(8) which
>> asks for the size when reading and manages it.  The second is
>> sysctl(2) where we have code like SCTL_MASK32 so a long on a 64-bit
>> kernel looks like a long to a 32-bit app.  It would be simpler to
>> expose this as an 8-byte quantity on 64-bit kernel, and sysctl(8) will
>> deal with this just fine, but that would break some uses of sysctl(2).
>
> What are the uses?  I guess they are not-so-hard-coding data types as
> int and long, etc., and hard-coding them as int32_t and int64_t.  Then
> you get an error when the kernel returns an unexpected length, but
> unexpected lengths are expected for 32-bit apps on 64-bit kernels.
>
> BTW, short writes are still horribly broken.  Suppose there is a size
> mismatch causing an application tries to write 32 bits to a 64-bit
> kernel variable.  Then no error is detected, and the kernel variable
> ois left with garbage in its top bits.  The error detection last worked
> with 4.4BSD sysctl, and is still documented to work:
>
> %      [EINVAL]           A non-null newp is given and its specified length
> in
> %                         newlen is too large or too small.
>
> I think the case where newlen is too large still works.
>
> I guess the bug is potentially larger for 32 bit compat applications,
> but it is rarely seen for those too since only system management
> applications should be writing to kernel variables and there is no
> reason to run such applications in 32 bit mode.

At Isilon all of userland is still 32-bit only.  The reason is that
for a while we supported 32-bit and 64-bit kernels in the field, and
in fact for various reasons on install the 32-bit kernel always came
up first and a reboot after install was forced for 64-bit capable
hardware.  But we needed the same userspace to work on both kernels.

>> However, I *think* it is safe to check the sysctl_req's oldidx/oldlen
>> to see what size the user-space expects and cast a kernel long to int
>> before SYSCTL_OUT.
>
> Maybe if the value fits.  The application might be probing for a size
> that works.  Then it shouldn't care what size the value is returned in,
> provided the value fits.  Writing is only slightly more delicate.  The
> application must use a size in which the value fits.  Then the kernel
> can expand the size if necessary.  It just has to be careful to fill
> the top bits and not have sign extension bugs.  sysctl_handle_i() can
> probably handle both directions.  Lower-level sysctl code should still
> return an error if there is a size mismatch.

Probing for a size that works should be passing in NULL for the old
ptr to get a hint, and for scalars the sie doesn't change.  It would
be a somewhat malformed app that probes for the size anywhere below
e.g. 512 bytes or 1 page.

> If a value doesn't fit, then the application will just have to detect
> the error and try a larger size (or fail if it doesn't support larger
> size).  An example might be a 32-bit app reading 64-bit kernel pointers.
> These probably have the high bits set, so they won't fit in 32-bit sizes.
> But the application can easily read them using 64-bit sizes.  Integers
> and pointers larger than 64 bits would be more interesting, since a
> compat application might not have any integer data type larger enough
> to represent them.
>
> So I changed my mind about converting in the kernel.  Just try to convert
> to whatever size the application is trying to use.  Don't even force >= 32
> bit sizes on applications.

The proposed code does attempt to convert to whatever the app uses,
but it assumes the app is using at least a 32-bit quantity. :-)

More on the other thread.

Thanks,
matthew
___
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"

svn commit: r217439 - stable/8/sbin/hastd

svn commit: r217440 - stable/8/lib/libc/net

svn commit: r217441 - stable/8/lib/libc/stdlib

Re: svn commit: r217369 - in head/sys: cam/scsi sys

svn commit: r217442 - stable/8/sys/kern

svn commit: r217443 - stable/7/sys/geom

svn commit: r217444 - head/sys/cam/ata

Re: svn commit: r217369 - in head/sys: cam/scsi sys

Re: svn commit: r217369 - in head/sys: cam/scsi sys

svn commit: r217445 - stable/8/sys/net

svn commit: r217446 - stable/8/usr.bin/netstat

svn commit: r217447 - stable/8/share/man/man9

svn commit: r217448 - stable/8/usr.bin/netstat

svn commit: r217449 - stable/8/sys/net

svn commit: r217450 - stable/8/sys/net

svn commit: r217451 - head/sys/powerpc/aim

svn commit: r217452 - head/sys/dev/iicbus

svn commit: r217453 - head/sys/vm

svn commit: r217454 - stable/8/lib/libkvm

svn commit: r217455 - stable/8/sys/net

svn commit: r217456 - stable/8/sys/net

svn commit: r217457 - stable/8/sys/net

svn commit: r217458 - stable/8/usr.bin/netstat

svn commit: r217459 - head/sys/powerpc/include

svn commit: r217461 - in head: bin/sh tools/regression/bin/sh/builtins

svn commit: r217463 - head/sys/vm

svn commit: r217464 - head/share/man/man4

Re: svn commit: r217369 - in head/sys: cam/scsi sys

svn commit: r217468 - head/share/man/man4

Re: svn commit: r217369 - in head/sys: cam/scsi sys

30 matches

Site Navigation

Mail list logo

Footer information