>So, tracing 32bit sys calls on 64 bit platform is the basic issue.
There are two sysent tables sysent (native, 64 bit) and sysent32(32 bit
system calls).
You need to modify the 32 bit and the 64 bit entry (and you need to provide
a native and a 32 bit implementation but in some cases they can
>I'm trying to create a pset with pset_create() and I get errno=1.
>In the man of pset_create() is written that this happens when the
>{PRIV_SYS_RES_CONFIG} privilege is not asserted in the effectvie
>set of the calling process.
>Is the problem indeed in the privilege?
truss will show that:
% t
>On Thu, 2010-11-25 at 00:55 -0800, Afantee Lee wrote:
>> Hi, All
>> With the help for my previous question, I checked out the source code tree.
>> of libc/port/thread
>> but, it seems that there is no way to compile pthread or libc only. there is
>> makefile under libc
, but, make all does not
>Hi Ian,
>
>yes volsize shows as 2.00G, and used as 2.13G and available 173Gb.
>
>But why does the system just die quietly ?
>
>I would expect some "noise" when a filesystem runs out of space,
>and with nothing else running (single user mode), would be expected ?
Does it die or does it hang?
H
>In theory, since rpool/swap is a zfs file system, it should be able to
>grow to the maximum free space of rpool, but maybe not ?
There's an issue with using zfs backed swap; the default configuration
needs to find memory in order to page out data and this creates an
additional memory pressur
>On Tue, 2010-10-12 at 17:12 +0200, casper@sun.com wrote:
>> >Do you like to tell us that the current code no longer supports Sun serial
>> >keyboards?
>>
>> The code is still there but the hardware is no longer supported.
>>
>>
>> I've actually added some code to the frkit "acpidrv" which
>Do you like to tell us that the current code no longer supports Sun serial
>keyboards?
The code is still there but the hardware is no longer supported.
I've actually added some code to the frkit "acpidrv" which can create a
panic or enter kdb when you hit the (soft) powerbutton three times qu
>Well, quite often you need to call time function in a loop. For
>example, you have a loop that accepts network packets, or connections
>and every so often you need to have current system time. The resolution
>of the timer can be specific to application, i.e. it could be seconds,
>milliseconds, o
>I just saw with truss the following:
>fstat64(5, 0x08062448) = 0
>fcntl(5, F_SETFD, 0x0001) = 0
>llseek(5, 1068220, SEEK_SET)= 1068220
>fstat64(5, 0x080452C0) = 0
>fstat64(5, 0x080451D0)
>I can't find man pages about flock and I can't compile my programs which uses
>flock(...).
>I found something similar called flockfile(). Does this mean flockfile should
>be used instead of flock?
No, flockfile() works on the inside a process; you should use fcntl()
locking (see fcntl(2) and lo
>Its worse than that -- unless the reference is in kernel memory, there
>is no reference on disk that goes backwards from inode to path name.
>find (see below) is about the only way that is guaranteed to work -- and
>of course that only finds inodes that actually *have* an active
>pathname.
>Thanks for holding my hand guys, this appears to work although I am certain it
>is ugly. Any direct suggestions ?
There is no guarantee that is returns the proper data all the time.
>Should I be locking anything while I am peeking around on the user stack?
>Perhaps "p_lock" ?
Won't help; th
>Hi All,
>
>I saw email many months ago about this, but now I can't find it. I want
>to get the source code matching specific builds of Nevada. So, for
>instance, the source for build 93. As I recall, there is at least one
>website which has all of the builds (well, at least, most of them).
>> There is only one way when the kernel knows and can
>> inspect the
>> environment: during evecve.
>
>Alright. Appears to be simply placing ptrs on the userstack, as an array,
>using the stk_putptr() method (based on sulword store). I can try to do that
>in reverse.
Correct. (It first copies
>I would like to see if a process (curthread->t_procp) has an environment
>variable $HOME, and if so, get the value of it.
>
>I know it is not something the kernel usually cares about (which probably
>explains why there are no examples of it) but I am testing a theory, and maybe
>learn a little mo
>Solaris 5.10 Generic_141445-09 i86pc
>
>I am playing around with a kernel module, learning some of the inner workings
>of the kernel, and I
would like to have a look at a process' environment.
>
>I am confused by the definition in user_t for environment:
>
> uintptr_t u_envp;
>
>.. since
>To avoid ambiguity: The problem is, that as far as I know the
>platform-specific versions of libc are still closed src binary-only
>releases. Otherwise we could of course just look at ON how they
>implemented the platform-specific libc-versions.
Have you looked in:
usr/src/lib/libc_psr?
Casper
>But not strcpy(), sprintf(), right?
What do you expect in strcpy/sprintf? Typically usage will use smallish
strings and the test of needing a *faster* strcpy/sprintf will typically
cost more for most cases.
Casper
___
opensolaris-code mailing list
op
>If there is a way to use VIS (or any other SPARC instructions for that
>mater) to make byteswapping faster then please implement them as sparc
>versions of hton[s,l,ll]() and fixup . Doing it that
>way means that ZFS, Crypto, Networking and a whole lot of other code
>gets to take advantage
>I've been tinkering with doors and I've hit a snag.
>
>When I create my door, the file file looks OK:
>
>did = door_create(doorFn, NULL, 0);
>
>fattach(did, "/tmp/demo");
>
>while( 1 ) sleep( 10 );
>
>Drw--- 1 ian staff 0 Mar 26 23:53 /tmp/demo
>
>But after I run a client that
>joerg.schill...@fokus.fraunhofer.de (Joerg Schilling) writes:
>
>> The problem with GCC-4.4 is that it cannot be called ANSI C compliant.
>> I did e.g. not jet find a way to prevent a warning for this line:
>>
>> (void)write(1, "Test" 4);
>>
>> which is 100% correct and even marks the return
>On Tue, Mar 2, 2010 at 7:02 AM, Garrett D'Amore wrote:
>>
>> I'm just thinking, isn't it time we just bit the bullet and told our users
>> to start using the POSIX defaults, instead of continuing to supply the
>> legacy compatibility stuff forever.
>>
>> It would save a little space, and maybe w
>Better is to use libast because its memory allocator allows switching
>between mmap(anon)/brk()/heap at runtime. Compile your application
>with -I/usr/include/ast -last and ln -s /usr/lib/libast.so.1
>/usr/lib/libast.so
Why? Do you have any roof that libast performs better then any other
memo
>Hi, I've read some reports that calling mmap() on SMP systems can be
>slow. Solaris has quite a few memory allocators, which makes a bit hard
>to figure out the best memory allocation technique on SMP hardware.
>
>I have single threaded application, which allocates large chunks of
>memory (256K a
>
>I have a program that fails on realloc. dbx reports:
>
>t...@1 (l...@1) signal SEGV (no mapping at the fault address) in t_splay at
>0xd078b710
>0xd078b710: t_splay+0x007e:movl %ebx,0x0008(%eax)
>Current function is _eina_strbuf_resize
> 521 buffer = realloc(buf->buf, new_size
>Bart Smaalders wrote:
>
>> All early strcpy implementations handle some overlapping copies
>> incorrectly. In order to determine safety, it is necessary to
>> find the length of the string before beginning the copy.
>
>Could you please give us an example where this typical implementation:
>
>ch
>My gut feeling is that the existing bug should be closed "not-a-defect"
>(if there is even a CR open for it), and Joerg (and any one else who has
>code that made bad assumptions) needs to fix their code.
>
>If we are terribly concerned that there are other programs from ISVs
>that have made t
>Previous versions of the POSIX standard made no limitations on strcpy()
>but the current standard claims that using strcpy() with overlapping
>buffers gives undefined results.
Even the Single UNIX ® Specification, Version 2, says:
If copying takes place between objects that overlap, the beh
>I just realised something, 'domainname' returns localnet.net, maybe this
>is what's appending .net to non-existent domain names. But then this
>doesn't explain why on occasions typing 'www.google.com' or any other
>valid name, takes web pages to other web servers??
>
That could just be a
>Solaris machine - IP: 192.168.1.1, Hostname: ultra10-hme0
>
>ISP DNS server - IP: 212.159.13.49
>
>
>I'm really confused, because when I run 'host' command on a non-existent
>domain, it returns that the domain is not found
>
>r...@ultra10 host -v www.somefakehost123.com
>Trying "www.somefakeho
>Hello Solaris developers, I need help with diagnosing possible malware,
>which seems to be originating from Solaris shared library.
>
>The problem occurs when I submit a URL that doesn't exist. Instead of
>returning an error, it seems Solaris resolver returns IP addresses of
>webserver whic
>
>>
>> Short form: if you're going to move stuff from the cloned
>> directory to a more handy build location, move *everything*
>> not just usr.
>>
>
>failures continue :
>
>http://www.blastwave.org/dclarke/OpenSolaris/nightly_06_Nov_2009_0905HRS_GMT.txt
>
>and this time I was careful to leave th
>Hi,
>
>I'm experiencing a problem that I'm trying to intercept the time series
>function. I was interested
when I noticed that open solaris is using a special fasttrap for fastening
time retrieve.
Does it work when you remove the cmn_err messages?
I'm sure that that will kill your system.
>Hi I'm getting the following crash in mutex_exit which is being done in the
>kernel.
>
>> ::status
>debugging crash dump vmcore.0 (32-bit) from solaris-devx
>operating system: 5.11 snv_72 (i86pc)
>panic message: mutex_exit: not owner, lp=d0eb3c94 owner=f6c4f510
>thread=d0eb3de0
>dump content:
>Oh yes that's what has happened, I checked the symbols with "nm driver2" and
>the driver1 symbols n
o longer appear as UNDEF. But that is not what I want.
>
>OK so here's what happens when I use your option of
>
>ld -dy -N /kernel/drv/driver1 -N /kernel/misc/scsi -r driver2.o -o driver2
You sh
>Yes that's correct, driver1 is present in /kernel/drv and driver2 has symbols
>it is using from dri
ver1 and also transport layer functions.
> The -Bdirect option seems to be working fine though.
Wel, if you actually used the command line as before:
>ld -dy -Bdirect driver1 -r driver2.
>Sorry Casper, small typo. Actually I wanted to use driver1 symbols in driver2,
>so what I'm using w
as actually
>ld -dy -N driver1 -N misc/scsi -r driver2.o -o driver2
Where is the driver1 installed? If it's installed in the standard
directories (/kernel/drv/...), then you should specify it a
>I've been using this linker line so far,
>ld -dy -N driver2 -r driver2.o -o driver2
>
>Upon your suggestion I modified this to
>ld -dy -N driver2 -N misc/scsi -r driver2.o -o driver2
The "-N driver2" is not correct. Are there other drivers you want to use?
"misc/scsi" is found using the stand
>Hi I have two loadable modules and one of them depends on the other for some
>symbols. I'm using
>ld -N driver1 -o driver2 driver2.o
>to resolve the symbol, but when I do so I'm unable to attach driver2 because
>it is unable to resol
ve symbols for scsi_hba_tran_alloc etc. and related API's ex
>So a situation at $WORK today has me wondering something - why is it
>that the pr_psargs member of psinfo is limited to 80 bytes (therefore,
>characters) ? Is this mainly due to historic reasons?
Well, /proc uses public structures; once they're defined, you cannot
change them.
But there ar
>Is it required to run rtc when I edit /etc/TIMEZONE file and change TZ
>variable ?
No; there's no direct connection between what's in /etc/rtc_config
and what is in /etc/TIMEZONE.
By default they are configured the same but you can also run the
rtc clock in a different timezone, e.g., GMT.
>Hi Guys,
>
>I have one x86 system installed with Solaris 10 Update 6. When I do
>TimeZone change , I have to run ntpdate command to synchronize
>the time of the system. I do following steps. My current time zone is
>in "Asia/Calcutta"
>
>1. Edit /etc/TIMEZONE and Change TZ variable to "Ameri
>When either adding a new class action script or updating an old one,
>how should this be tested?
>
>Obviously bfu doesn't test this.
"acr" does, I think.
>Is it possible to do a limited upgrade install using the bits from nightly?
Well, typically class action scripts are run as follows:
ech
>Extended Accounting stores project ids only. Not having project ids
>consistent across system will make generating accounting reports across
>systems painful since one would need to resolve the project id in the
>accounting file to the project name on the originating system, instead
>of on t
>For automaticly loaded modules, there is always something that pulls the
>module into the kernel, be it a dependency from another module or
>someone opening a device file. Neither of those is going to pull in a
>new standalone module, so that code can be executed.
How many of those do we hav
>Alan Coopersmith wrote:
>> Is there a public interface to load a kernel module from user space?
>>
>> Looking at the modload source, it seems to use the modctl() system call
>> but I see no man page for that.
>>
>> (At this point it's a matter of curiosity, since I noticed that the
>> Xorg code
>A year and a half since I've had my Toshiba Satellite 305D .. and STILL
>no support for Solaris 10 .. just sad.
>
>If anyone has any idea when this will work, please let me know if it's
>on the radar.. or whom I cann contact on the cert team to get some
>insight (or work-around drivers..??).
>Hi,
>
>Recently we hit a problem with one of our file systems where 2 kernel threads
>were killed holding
a kmutex. Due to this all the applications which use the fs blocked waiting for
the kmutex.
>
>I am wondering why there is no synchronization type equivalent to
>USYNC_PROCESS_ROBUST (whi
>Hello Darren,
>
>Wednesday, April 1, 2009, 2:47:31 PM, you wrote:
>
>DJM> Robert Milkowski wrote:
>>> It would also require adjustment of setppriv() at
>>> http://src.opensolaris.org/source/xref/onnv/o
nnv-gate/usr/src/uts/common/syscall/ppriv.c#57
>>>
>>> and perhaps somewhere else.
>>>
>>> B
>In my perfect privilege system we would have all of P,I,E,L,F,A. Now
>that OpenSolaris has moved to ZFS root and we have a more complete
>system/xattr system we could reintroduced the A and F sets.
But not in a process; since we're talking about the process model, F&A
aren't important.
Casp
>
>
>>The above code change would allow growing L set if new L' set is a subset
>>of the effective set of the calling process.
>>What would be the impact? Would it break anything?
>
>
>The limit has been designed as a limit you cannot escape.
>
>Changing that would break a promise the privilege sy
>The above code change would allow growing L set if new L' set is a subset
>of the effective set of the calling process.
>What would be the impact? Would it break anything?
The limit has been designed as a limit you cannot escape.
Changing that would break a promise the privilege system makes.
>"I. Szczesniak" wrote:
>
>> Then search again, please. We were unable to port two products to
>> Solaris because Sun only implemented the minimum number required by
>> POSIX (8) that Sun marketing can add the 'certified POSIX-conforman=
>t'
>> sticker.
>
>Solaris _is_ POSIX compliant as Solaris
I'm not sure how brandz manipulates signals, but if the RFE also requires
to change the RT signals in Solaris, then that is fine.
Perhaps they only use brandz because Solaris doesn't support sufficient
RT signals?
Casper
___
opensolaris-code mailin
>64 realtime signals would be OK for a 64 bit operating system.
Why would that need to be different for a 64 bit OS?
I wonder why people use SIGTRXXX signals or signals at all.
Casper
___
opensolaris-code mailing list
opensolaris-code@opensolaris.o
>On 3/23/09, James Litchfield wrote:
>> To make it clear - there is no magic tunable to change the number
>> of RT signals. It requires changing source (catching every place the
>> current data structures used are too small) and rebuilding.
>
>Why does Solaris only support 8 signals? AIX suppor
>ff0192256020 19204 ff01ac214558 cralloc_flags+0x1e
>ff0192256020 1902833 ff01ac212db8 cralloc_flags+0x1e
>ff019131fb20 1921659 ff0192255bf8 labelalloc+0x2f
These are pretty horrible. Try the ::bufctl comand in mdb.
You're using Trusted Extensions?
Casper
>Hi,
>
>I'm debugging an issue with GNU tar. Without going into a lot of
>details (though I can if needed), here's a truss snippet of 'tar -xzf
>foo.tar.gz' running as root:
>
>2655: getppriv(PRIV_EFFECTIVE, {}) = 0
>2655: all
>2655: setppriv(PRIV_SET, PRIV_EFFECT
>http://askdrroot.blogspot.com/2006/10/q-can-i-run-linux-applications-on.html
Did you install a linux zone, then?
Casper
___
opensolaris-code mailing list
opensolaris-code@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/opensolaris-code
>I hear that i can use all the software for linux in opensolaris, i tried to
>install a package and it did not work, how can do it?
Who told you that?
"From the source" you need to compile the software.
Casper
___
opensolaris-code mailing list
opens
>A customer had a small program using pthreads that crashed pretty much at
startup. After verifying the crash on my system, I noticed that the customer
>had not linked the program with libpthread. Adding -lpthread to the linking
> command fixed the problem.
>
>If a program does not link a needed l
>casper@sun.com wrote:
>
>>
>> >Hi,
>> >
>> >I add some test codes to ../uts/common/os/kcpc.c.
>> >
>> >int
>> >kcpc_bind_thread(kcpc_set_t *set, kthread_t *t, int *subcode)
>> >{
>> >..
>> >/* new codes. */
>> >cmn_err(CE_NOTE, "ks_data = %" PRId64 ".\n", (uint64_t)set->
>Hi,
>
>I add some test codes to ../uts/common/os/kcpc.c.
>
>int
>kcpc_bind_thread(kcpc_set_t *set, kthread_t *t, int *subcode)
>{
>..
>/* new codes. */
>cmn_err(CE_NOTE, "ks_data = %" PRId64 ".\n", (uint64_t)set->ks_data);
>..
>}
You're printing a pointer; you need
>There does have the requirement to read files in the kernel.
>
>We're working on a 10Gb Ethernet driver and we want to let the driver
>update the firmware on-the-fly if needed when the driver is loaded.
>Save the firmware image in a separate file and let the driver read it
>with kobj_open_file()
>On Jan 27, 2009, at 6:03 PM, Steve Scargall wrote:
>
>> I doubt there's many people on this alias that will be able to
>> answer your question with any great detail. I'm no OBP expert but
>> have played around with it at times especially for debugging low-
>> level hardware issues. It's ex
>Is the premature return from the system call (not caused by signal
>delivery) allowed by any standard? How is this case handled in other
>operating systems?
>
>E.g. the www.opengroup.org page for poll() tells me
>that "... poll() shall wait at least timeout milliseconds
>for an event to occur on
>The putback for 6565503 "callout processing is single threaded, throttling
>applications that rely on scalable callouts" in build 103 apparently has
>changed the kernel function cv_waituntil_sig() to create callouts with
>a new flag CALLOUT_FLAG_HRESTIME. This flag is described in
>uts/common/sy
>On Tue, Jan 13, 2009 at 5:12 AM, wrote:
>>
>> I see no reason for "access(file, X_OK))" should return 0 for
>> ordinary files which have no execute bit set.
>>
>> I think we can just change that.
>
>This behaviour is controlled by the file_dac_execute privilege.
>Processes running as root will
>Erm... I doubt it's "just 10 lines" - you need a bit more code. Beyond
>that we would have to maintain two codelines, including for the
>following upcoming changes:
>- Adding "-s" for alternative signal handling
>- Use of |sched_yield()| for zero values (based on empirical testing for
>the built
>from /proc/*/psinfo:
>* pr_psargs (first 80 characters of args as of exec time)
>* pr_fname (exec name)
>* pr_argv (offsets into process address space?)
Correct; and you need to have access to the process memory before you
can dereference them. (ps is no longer set-uid root; even ps, then, h
>On Tue, Jan 13, 2009 at 2:20 PM, Joerg Schilling
> wrote:
>> casper@sun.com wrote:
>>
>>>
>>> >
>>> > > I see no reason for "access(file, X_OK))" should return 0 for
>>> > > ordinary files which have no execute bit set.
>>> > >
>>> > > I think we can just change that.
>>> >
>>> >Modulo due di
>
> > I see no reason for "access(file, X_OK))" should return 0 for
> > ordinary files which have no execute bit set.
> >
> > I think we can just change that.
>
>Modulo due diligence to see what in our codebase might be depending on it.
>Is it really worth it?
As supposed to adding __posix_acces
I see no reason for "access(file, X_OK))" should return 0 for
ordinary files which have no execute bit set.
I think we can just change that.
Casper
___
opensolaris-code mailing list
opensolaris-code@opensolaris.org
http://mail.opensolaris.org/mailma
>Is there a way that multiple proceses can write to the same pipe fd but
>keep their writes "atomic" (regardless of the size, e.g. the writes may
>be 1 byte or 300GB), e.g. that the data (or better "record") being
>written appears in one piece ?
Is it a problem if each process gets their own pip
>When compiling the Solaris AVL and zfs DMU modules on Windows I get many
>warnings where a uintptr_t is cast to an int, and vice versa. The actual
>cases seem benign (e.g. AVL is doing unitptr_t & 1 -> int, which is OK) but
>I'm worried there might be some real issues
>Just for my sanity, is "i
>> Please do not bother - use the glm(7D) driver instead.
>
>I thought glm only worked with Sun branded controllers.
>I was never able to make glm recognize a Tekram controller
>with the 53c896 chip that I own.
>It would be nice to be able to use that controller again.
And you didn't use the (o
>Yepp - seems that clamav-milter needs a fork anyway ;-)
>However I didn't know, that a "simple" malloc could result into a SEGV -
>assumed, that the OS core takes care of managing memory and prevents
>apps from accessing/destroying its tables ...
There's no protection between "libc memory" and
>It seems bogus, but if I recall correctly, the standards conformance
>gurus say that we're not allowed to walk into the user's name space in
>this way: those symbols are defined by the standards, so they must not
>exist. If you don't like that, then don't compile in a standards-
>conformant env
>kevin writes:
>> In /usr/include/sys/dirent.h I see this:
>>
>> /*
>> * File-system independent directory entry.
>> */
>> typedef struct dirent {
>> ino_t d_ino; /* "inode number" of entry */
>> off_t d_off; /* offset of d
>I am trying to port a linux app to OpenSolaris OS. There is a API used in
>linux codes that I donât know how to port to OpenSolaris:
>ioctl(fileFD, FIBMAP, &FsBlock)
>Any one pleases help, tell me how to do this, find out a FileSystemBlock of the
offset in a file, in OpenSolaris.
You can't.
>Right now, I have a driver (the audio subsystem I'm working on) which
>will ultimately want to be integrated into Nevada. But it wants to
>access some fields in the uarea of the process that opened the driver
>initially -- specifically the command and arguments.
>
>My code currently does this
I'm surprised that "sun" is still defined; clearly that is true ONLY when
you run in K&R mode?
Casper
___
opensolaris-code mailing list
opensolaris-code@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/opensolaris-code
>Hi Casper,
>
>On Mon, Oct 6, 2008 at 10:28 PM, <[EMAIL PROTECTED]> wrote:
>>
>> The system should not produce scaring messages when there is no action to
>> be taken.
>>
>> (I see the same message on an old P3; we don't support any form of
>> speedstep on that)
>>
>
>These log messages are helpf
>On Mon, Oct 06, 2008 at 09:35:19AM +0200, [EMAIL PROTECTED] wrote:
>>
>> >T-state is Processor Clock Throttling state. It is designated to throttle
>> >the
>> >processor clock in a few specific situations, like the CPU temperature
>> >exceeds a threshold value.
>> >
>> >T-state is defined in AC
>T-state is Processor Clock Throttling state. It is designated to throttle the
>processor clock in a few specific situations, like the CPU temperature
>exceeds a threshold value.
>
>T-state is defined in ACPI spec and need BIOS implementation to get
>supported. The warning message means T-state i
>Hi all,
>
>I am currently porting a driver from Linux to Solaris. The driver code
>makes use of the likely macro that is defined on Linux to tell the
>compiler that the code inside the likely call is the most likely code
>path to optimize it.
>
>Does someone know if it is available on Solaris
>Hi
>
>Thanx for the quick answer..
>
>> The fix is using the new "F" mode character which tells fopen that it is
>> fine to open files with a file descriptor > 255.
>>
>> The actual fix is:
>>
>>- add support for "F" etc in fopen()
>>- add "F" when it is needed (basically everythi
>Hi
>
>I have found these two bug reports regarding the fopen usage which both
>have the status fix-delivered.
>
>http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6221296
>http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6234782
>
>
>however when I browse the source code using
>Related to that, I've found a few header files where you must include
>a before including b -- essentially 'b' requires definitions in 'a',
>but itself does not include 'a'. I'd have to go back to find which
>ones specifically (I found them doing some porting of a network
>driver). Should such
>> Currently Solaris has two distinct versions of ps; to use one or the other
>> you need to either give the full path or change $PATH.
>>
>> And that while a little bit of code could tell most invocations apart.
>>
>> Because it has been bothering me I'd like to fix this and so I'd like to
>> ask
>On Fri, Feb 15, 2008 at 7:30 PM, <[EMAIL PROTECTED]> wrote:
>>
>> But if you type /usr/ucb/ps or /usr/bin/ps, what do you expect?
>>
>> Or PATH=/usr/ucb:$PATH ps
>> vs
>>PATH=/usr/bin:$PATH ps
>
>If arguments start with a -, be /usr/bin/ps; otherwise be
>/usr/ucb/ps, in all cases.
>On Fri, Feb 15, 2008 at 6:58 PM, Kyle McDonald <[EMAIL PROTECTED]> wrote:
>
>> >
>> I'm not sure, but I read this as:
>>
>> if argv[0]="/usr/ucb/ps" then interpret the args the ucb way even if
>> there is a -.
>>
>> That doesn't sound bad to me. If people put that in their path that's
>> pr
>
>> From [EMAIL PROTECTED] Fri Feb 15 11:05:28 2008
>> >
>> >The other question would be - which output do you get with no
>> >arguments at all?
>>
>> The same as currently, depending on $PATH. I'm not sure what other
>> compatible options there are.
>
>Are you parsing $PATH when argv[0]=="ps"
>I'm not sure, but I read this as:
>
>if argv[0]="/usr/ucb/ps" then interpret the args the ucb way even if
>there is a -.
Correct.
>That doesn't sound bad to me. If people put that in their path that's
>probably what they expect.
That was my thinking too.
Casper
___
>On Fri, Feb 15, 2008 at 1:55 PM, <[EMAIL PROTECTED]> wrote:
>>
>>
>> Currently Solaris has two distinct versions of ps; to use one or the other
>> you need to either give the full path or change $PATH.
>>
>> And that while a little bit of code could tell most invocations apart.
>>
>> Because
> ENVIRONMENT VARIABLES
> The following environment variables could affect ps:
>
> COLUMNS Override default display width
> LINESOverride default display height
> PS_PERSONALITY Set to one of posix,old,linux,bsd,sun,digital
> CMD_
Currently Solaris has two distinct versions of ps; to use one or the other
you need to either give the full path or change $PATH.
And that while a little bit of code could tell most invocations apart.
Because it has been bothering me I'd like to fix this and so I'd like to
ask some advice.
Mer
>How would you do this on OpenSolaris (x86/Sparc) ?
>With a second machine ?
You don't need a second machine
Just boot with the kernel debugger enabled (w/ the -k option)
or run "mdb -K" from the commandline console.
Casper
___
opensolaris-code mail
>It's failing in the checking of the security policy, specifically:
>
>boolean_t
>priv_isfullset(const priv_set_t *set)
>{
>PRIV_TEST_BODY(set->pbits[i] == ~(priv_chunk_t)0);
>}
>
>Evidently I'm missing something from my hand assembled system that defines the
>policies. Any clues
as to w
>Actually, as I look around more, I see that some other systems
>(*BSD, Linux, QNX) have one or both of lchmod(), flink(). One
>argument used for both even if they otherwise seem silly is avoidance
>of race conditions, although I've seen counter-arguments for flink(),
>but I didn't quite understa
1 - 100 of 302 matches
Mail list logo
