user-mode-linux-devel

[User-mode-linux-devel] Re: Problems compiling 2.4.0test2 patch...

[Jeff D. <jd...@ka...>]

sa...@sk... said:
> Okay, I just configured it with smp support and built.  Not much to go
> wrong. 

Oh.  You didn't mention that you configured it with SMP.

> perhaps then you should remove that option from the configure process,

Maybe.  But it's been there long enough and I'm close enough to getting around 
to making it work again, that I think I'll just leave it there.

> does this mean that all the locks are noops?  Is that really what you
> want?

For a UP kernel, all the locks are noops, which is fine.

> Gee, sorry to try to help. 

Like I said, it sounded like you ran into some compilation error that involved 
some __SMP__ code somehow and you decided to turn on -D__SMP__.

So going back to the problems you found:

> Warning: unable to open an initial console. 

I don't know what this is about.

> Seg fault in signals 

This I can make a better guess about.

Under SMP, a problem you have to deal with is getting a process's 
current_task.  Linux does it by putting the task structure at the bottom of 
the process kernel stack so it can get its current_task by zeroing the bottom 
11 bits of its sp.

Sometimes something needs current_task, but isn't running on the kernel stack. 
 This is true of the tracing thread, which needs the current_task of whatever 
process got a signal, and sometimes true of the process itself, which can run 
kernel code on the process stack when delivering signals.

So, there's an array, cpu_tasks which maps pids (in the host kernel) to task 
structures.  The tracing thread looks the task structure up there.

I just had a look at restore_state.  It's only called from exit_kernel, which 
means that it needs to pass in task rather than NULL.

That will fix that problem.  That #error that I put in do_bh needs to be 
looked at, otherwise softirqs probably won't work.

				Jeff

[User-mode-linux-devel] Re: Problems compiling 2.4.0test2 patch...

[Chris L. <sa...@sk...>]

> sa...@sk... said:
> > FYI, for some reason, I can't compile 2.4.0 out of the box (the UM
> > patch that is...).  It seems that you are doing a bunch of #ifdef
> > __SMP__'s, but __SMP__ is not defined... 
> 
> I just did a build of the patch applied to a clean 2.4.0-test2 pool, and it 
> built fine.  Something's wrong at your end.

Okay, I just configured it with smp support and built.  Not much to go
wrong.

> > I added -D__SMP__ to my command line and things seem quite a bit
> > happier now.  :) 
> I have SMP support turned off for now, with pretty good reason.

okay... perhaps then you should remove that option from the configure
process, does this mean that all the locks are noops?  Is that really what
you want?

> If you want to get it to build, get a clean pool, and it will probably work.  
> If it doesn't, send in the compile errors.  Don't fix non-existent problems.
> 
> If you want to get SMP support working, let me know, and I'll be happy to 
> supply hints.  There is more to SMP support than removing #errors and putting 
> in kludges to make it build.

Gee, sorry to try to help.

-Chris

[User-mode-linux-devel] Re: Problems compiling 2.4.0test2 patch...

[Jeff D. <jd...@ka...>]

sa...@sk... said:
> FYI, for some reason, I can't compile 2.4.0 out of the box (the UM
> patch that is...).  It seems that you are doing a bunch of #ifdef
> __SMP__'s, but __SMP__ is not defined... 

I just did a build of the patch applied to a clean 2.4.0-test2 pool, and it 
built fine.  Something's wrong at your end.

> I added -D__SMP__ to my command line and things seem quite a bit
> happier now.  :) 

I have SMP support turned off for now, with pretty good reason.

> --- arch/um/kernel/process_kern.c-  Mon Jul  3 05:05:03 2000 
> +++ arch/um/kernel/process_kern.c       Mon Jul  3 10:34:23 2000 
> @@ -228,7 +228,9 @@
>         if (softirq_state[0].active&softirq_state[0].mask)
>                 do_softirq();
>  #else 
> -#error Need to update do_bh 
> +/* 
> + * #error Need to update do_bh 
> + */
>  #endif
>  } 

This being one of them.

> In spite of making these changes I'm still getting this error:
> VFS: Mounted root (ext2 filesystem) readonly. 
> Warning: unable to open an initial console. 
> Seg fault in signals 

Which is another reason I have SMP turned off.  

If you want to get it to build, get a clean pool, and it will probably work.  
If it doesn't, send in the compile errors.  Don't fix non-existent problems.

If you want to get SMP support working, let me know, and I'll be happy to 
supply hints.  There is more to SMP support than removing #errors and putting 
in kludges to make it build.

				Jeff

[User-mode-linux-devel] [PATCH] Compile cleanly with SMP support...

[Chris L. <sa...@sk...>]

Okay, here's a patch that should make it compile happier with SMP.  It
makes the following changes:

1. Define the __SMP__ macro [bad, shouldn't go in top level makefile]
2. Converting the cpu_number_map array to a macro so that sched.c will
   compile...
3. #include <linux/sched.h> from asm/smp.h so that dereferencing current
   works right...
4. Comment out #error in do_bh
5. #ifdef out some stuff in irq.c to fix undefined symbol
   refs: apic_timer_irqs and cpu_online_map.
6. Silence some undefined function name warnings
7. #ifndef CONFIG_SMP around EXPORT_SYMBOL(current_task);

The "Makefile" change is not optimal (ie it would be better to change all
the __SMP__s to CONFIG_SMP throughout the source) but this should work.

In spite of making these changes I'm still getting this error:

VFS: Mounted root (ext2 filesystem) readonly.
Warning: unable to open an initial console.
Seg fault in signals

Which I'm having problems decifering.  It looks like an illegal address
for a task_struct in restore_state (process_kern.c).

This may be fixed in one of your patches, although I applied both patches
listed on your Dear Diary page
(http://user-mode-linux.sourceforge.net/diary.html) and they didn't
help...

Besides those two patches, this is based on the 2.4.0test2 patch.

Help! :)

-Chris

--- Makefile-   Mon Jul  3 04:44:05 2000
+++ Makefile    Mon Jul  3 04:30:57 2000
@@ -85,7 +85,7 @@
 
 CPPFLAGS := -D__KERNEL__ -I$(HPATH)
 
-CFLAGS := $(CPPFLAGS) -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer
+CFLAGS := $(CPPFLAGS) -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer
-D__SMP__
 AFLAGS := -D__ASSEMBLY__ $(CPPFLAGS)
 
 # use '-fno-strict-aliasing', but only if the compiler can take it


--- include/asm-um/smp.h-       Mon Jul  3 04:35:41 2000
+++ include/asm-um/smp.h        Mon Jul  3 04:48:46 2000
@@ -2,6 +2,7 @@
 #define __UM_SMP_H
 
 #include "asm/current.h"
+#include "linux/sched.h"    /* Get the definition of task_struct */
 
 #ifdef __SMP__
 
@@ -9,7 +10,8 @@
 
 #define cpu_logical_map(n) (n)
 
-extern int cpu_number_map[];
+#define cpu_number_map(x) (cpu_number_map_a[x])
+extern int cpu_number_map_a[];
 
 #define PROC_CHANGE_PENALTY    15 /* Pick a number, any number */


--- arch/um/kernel/smp.c-       Mon Jul  3 04:39:11 2000
+++ arch/um/kernel/smp.c        Mon Jul  3 04:39:05 2000
@@ -26,7 +26,7 @@
 struct cpuinfo_um cpu_data[NR_CPUS];
 
 /* which CPU maps to which logical number */
-int cpu_number_map[NR_CPUS];
+int cpu_number_map_a[NR_CPUS];
 
 spinlock_t um_bh_lock = SPIN_LOCK_UNLOCKED;
 
--- arch/um/kernel/process_kern.c-      Mon Jul  3 05:05:03 2000
+++ arch/um/kernel/process_kern.c       Mon Jul  3 10:34:23 2000
@@ -228,7 +228,9 @@
        if (softirq_state[0].active&softirq_state[0].mask)
                do_softirq();
 #else
-#error Need to update do_bh
+/*
+ * #error Need to update do_bh
+ */
 #endif
 }


--- arch/um/kernel/irq.c-       Mon Jul  3 05:00:37 2000
+++ arch/um/kernel/irq.c        Mon Jul  3 05:00:19 2000
@@ -117,7 +117,7 @@
                p += sprintf(p, "%10u ",
                        atomic_read(&nmi_counter(cpu_logical_map(j))));
        p += sprintf(p, "\n");
-#if CONFIG_SMP
+#if CONFIG_SMP_x
        p += sprintf(p, "LOC: ");
        for (j = 0; j < smp_num_cpus; j++)
                p += sprintf(p, "%10u ",
@@ -609,7 +609,7 @@
 
        err = parse_hex_value(buffer, count, &new_value);
 
-#if CONFIG_SMP
+#if CONFIG_SMP_x
        /*
         * Do not allow disabling IRQs completely - it's a too easy
         * way to make the system unusable accidentally :-) At least


--- arch/um/kernel/syscall_kern.c-      Mon Jul  3 05:01:32 2000
+++ arch/um/kernel/syscall_kern.c       Mon Jul  3 05:03:08 2000
@@ -18,6 +18,9 @@
 #include "kern_util.h"
 #include "user_util.h"
 
+long sys_mount(char * dev_name, char * dir_name, char * type,
+                          unsigned long new_flags, void * data);
+
 long um_mount(char * dev_name, char * dir_name, char * type,
              unsigned long new_flags, void * data)
 {
@@ -186,6 +189,9 @@
        fd_set *inp, *outp, *exp;
        struct timeval *tvp;
 };
+
+asmlinkage long
+sys_select(int n, fd_set *inp, fd_set *outp, fd_set *exp, struct timeval *tvp);
 
 int old_select(struct sel_arg_struct *arg)
 {


--- arch/um/kernel/um_arch.c-   Mon Jul  3 05:03:53 2000
+++ arch/um/kernel/um_arch.c    Mon Jul  3 05:06:04 2000
@@ -147,8 +147,10 @@
                        physmem_size = memparse(argv[i] + strlen("mem="), 
                                                &retptr);
 #ifdef __SMP__
-               else if(!strncmp(argv[i], "ncpus=", strlen("ncpus=")))
+               else if(!strncmp(argv[i], "ncpus=", strlen("ncpus="))) {
+                       extern int strtoul(const char*, void *, int);
                        ncpus = strtoul(argv[i] + strlen("ncpus="), NULL,
10);
+           }
 #endif
                else if(!strncmp(argv[i], "debugtrace", strlen("debugtrace")))
                        debug_trace = 1;

--- arch/um/kernel/init_task.c- Mon Jul  3 05:35:34 2000
+++ arch/um/kernel/init_task.c  Mon Jul  3 05:35:45 2000
@@ -10,7 +10,9 @@
 #include "asm/pgtable.h"
 #include "user_util.h"
 
+#ifndef CONFIG_SMP
 EXPORT_SYMBOL(current_task);
+#endif
 
 static struct vm_area_struct init_mmap = INIT_MMAP;
 static struct fs_struct init_fs = INIT_FS;

[User-mode-linux-devel] Problems compiling 2.4.0test2 patch...

[Chris L. <sa...@sk...>]

FYI, for some reason, I can't compile 2.4.0 out of the box (the UM patch
that is...).  It seems that you are doing a bunch of #ifdef __SMP__'s, but
__SMP__ is not defined...

I think the symbol you want to use is CONFIG_SMP?

I added -D__SMP__ to my command line and things seem quite a bit happier
now.  :)

-Chris

[User-mode-linux-devel] CVS updates

[Jeff D. <jd...@ka...>]

I checked in the fixes for the two problems that cropped up last week.  One 
was caused by the increased signal frame size in recent kernels, so if you are 
trying and failing to run uml on a new kernel, this is fixed.  The other one 
was caused by a stacksize limit not being a multiple of 4 meg and inducing uml 
to unmap the main process stack.  That caused a panic when the process tried 
to update its name.

For helping me diagnose that last problem, Gord Lamb gets an entry on my 
thanks page.

				Jeff

[User-mode-linux-devel] Recognition

[Jeff D. <jd...@ka...>]

I added a page (http://user-mode-linux.sourceforge.net/thanks.html) to my site 
thanking people who have contributed to the project.  If you are one of the 
following people, have a look and see if you are happy with your entry:

Rusty Russell 
Peter Moulder
Bill Stearns
Yuri Pudgorodsky 
Gerald Britton 
Ian Wehrman

Let me know if you want any changes made if you're not.

If you're not listed and you think you should be, let me know.  

The criteria I used is that if what you did is in and working in whatever way 
makes sense, then you should be listed.  There are a few things which are in 
progress which will get listed when they work.

I also added a diary page which I may keep somewhat up to date with what's 
going on.  Or I may not.

				Jeff

Re: [User-mode-linux-devel] uml at ALS2000

[Jeff D. <jd...@ka...>]

la...@no... said:
> It's also very appropriate that he's the one to make sure deadlines
> are met. :) 

OK, so I missed the RTLinux reference... :-)

				Jeff

Re: [User-mode-linux-devel] uml at ALS2000

[Jeff D. <jd...@ka...>]

la...@no... said:
> Jeff Dike <jd...@ka...> writes:
> > That makes it sound like Victor is going to come over here and break
> > fingers if I don't make the deadline...

> It's also very appropriate that he's the one to make sure deadlines
> are met. :) 

Why?  Do you know something about him that I don't?

Should I plan on moving and changing my name if I don't meet the deadlines :-)

				Jeff

Re: [User-mode-linux-devel] uml at ALS2000

[lars b. <la...@no...>]

Jeff Dike <jd...@ka...> writes:
> That makes it sound like Victor is going to come over here and break fingers 
> if I don't make the deadline...

It's also very appropriate that he's the one to make sure deadlines are
met. :)

[User-mode-linux-devel] uml at ALS2000

[Jeff D. <jd...@ka...>]

I submitted an abstract to the Atlanta Symposium a while back and it's been 
accepted.  So, if any of you are going to be there, MAKE SURE YOU ATTEND THIS 
VITALLY IMPORTANT TALK :-)

There is a slightly humorous part of the email I got from Ted T'so:

> Each paper has been assigned a member of the program committee to act
> as a shepherd [ who is Victor Yodaiken - jd ]. The role of the shepherd is to
> ...
> ensure that you produce an appropriate paper by the August 24th
> deadline.

That makes it sound like Victor is going to come over here and break fingers 
if I don't make the deadline...

				Jeff

[User-mode-linux-devel] 2.4.0-test2 checkins

[Jeff D. <jd...@ka...>]

I've updated CVS to 2.4.0-test2.  No major changes in this one, just a devfs 
interface change or two and ptrace flags getting split out from the other 
task_struct flags.

The one functional change is that I added support in the console driver to 
grab a pty without starting up an xterm.  This is useful when running uml 
while ssh'd to a remote machine which isn't allowing X to be tunnelled.

				Jeff

[User-mode-linux-devel] Toms Root/Boot and 2.3.99, v2

[William S. <wst...@po...>]

Good evening, Jeff,
	The current toms and your new 2.3.99pre9 look moderately good
together!  Nice work!
	The work isn't done, but could we seriously look at getting toms
up on Sourceforge as an early release for others to play with?
	Cheers,
	- Bill

---------------------------------------------------------------------------
	"Eagles may soar, high and proud, but weasels don't get sucked
into jet engines."
(Courtesy of Mike Andrews <man...@te...>)
--------------------------------------------------------------------------
William Stearns (wst...@po...).  Mason, Buildkernel, named2hosts, 
and ipfwadm2ipchains are at:                http://www.pobox.com/~wstearns
LinuxMonth; articles for Linux Enthusiasts! http://www.linuxmonth.com
--------------------------------------------------------------------------

[User-mode-linux-devel] Another batch

[Jeff D. <jd...@ka...>]

One more set of checkins before the next release.  This one makes the ubd 
driver fall back to synchronous I/O if it can't start the I/O thread.  The 
main reason this can happen is that the I/O thread communicates with the 
kernel with a pts pty, which can't be allocated by a normal user on sme 
distributions.  So, not having a backup plan would make uml unusable for those 
people without running as root.

I also changed arch/um/defconfig and bumped EXTRAVERSION in arch/um/Makefile.

				Jeff

[User-mode-linux-devel] more checkins

[Jeff D. <jd...@ka...>]

This batch is bug fixes:

the IRQ handlers now run with interrupts off
fixed a bug which could caused the same interrupt to effectively be handled 
twice
SIGALRM is treated the same as SIGVTALRM
the old select system call now works

and the big item:

the occasional process segfaults are now fixed (I think).  They became a lot 
more reproducible over the last week, so I chased the bug down.  The several 
days that I spent looking at the debris left behind by the bug were useless.  
What ended up happening was that I thought about it long enough to realize 
what the bug must be and tossed in the line of code that fixed it.

				Jeff

Re: [User-mode-linux-devel] New root_fs; problems with Toms Root/boot (libc5)

[Jeff D. <jd...@ka...>]

wst...@po... said:
> 	I get the impression I didn't add those devices correctly; I added /
> dev/udbX instead of /dev/ubd/X.  Which is correct? 

The convention would be /dev/udbX in a non-devfs environment, but it's 
arbitrary, since you're going to have to stick those names in fstab (which is 
about the only thing that cares), and mount will use whatever names you give 
it.

> 	I'll lay odds that was it.  I take it this is a global fix that
> needed to be made, rather than some custom fix for Toms?

Yeah.  It was stupid, and I thought I had checked all of the entry points to 
make sure they were right.

> It actually does create /etc/passwd when the system boots.  From /fl/
> rc.custom.gz:

Then why didn't it?

> As a side note, what would you think of splitting the kernel from the
> root_fs instead of making them a unified package?  That would save you
> the effort of rebuilding the unified package each time a new kernel
> comes out.  On the other hand, it means an extra step for the users.
> 6 of one, half dozen of the other...

I do that as a convenience.  It occupies an interesting (NOT) hour or so of my 
time fiddling with the sourceforge webby interface.  So, it's not all that big 
a deal.

				Jeff

Re: [User-mode-linux-devel] New root_fs; problems with Toms Root/boot (libc5)

[William S. <wst...@po...>]

Good afternoon, Jeff,

On Sun, 18 Jun 2000, Jeff Dike wrote:

> I got it to boot with no changes to the disk itself except for adding an 
> /etc/passwd (not even /dev/ubd/*, which I forgot :-).

	I get the impression I didn't add those devices correctly; I added
/dev/udbX instead of /dev/ubd/X.  Which is correct?

> Your immediate problem is probably that I have the old select syscall hooked 
> up wrong.  It goes into an infinite loop returning an error because the 
> arguments are being picked up wrong in the kernel.  Fixing that and another 
> problem gives me a clean boot.

	I'll lay odds that was it.  I take it this is a global fix that
needed to be made, rather than some custom fix for Toms?

> If you are building from sources, I can give you patches which fix things.

	I should start doing that someday.  *smile*  Darn it, it's so
_convenient_ to use your precompiled ones!  On the other hand, I would
like to see if the uml patch meshes with the ac-x series.

> And what's with no /etc/passwd, anyway?  Is there some other way to get a 
> shell out of it?

	It actually does create /etc/passwd when the system boots.  From
/fl/rc.custom.gz:
cat>passwd<<X
root:$PASSWD:0:0::/:/bin/sh
shutdown:*:6:0::/:/usr/bin/shutdown
halt:*:7:0::/:/usr/bin/halt
X
	and /fl/settings.s:
PASSWD=xxxx

> > I'd love to see it as one of the available root environments for uml.
> > He does have a short copyright message and two urls that he'd like
> > included in any derivative works, but I suspect that's not too much of
> > a problem; they could be included in a short README.toms.
> 
> OK, I can stick it up there once I put up a kernel that can boot it.

	Deal.  If it makes sense to you, we could have both the prebuilt
root_fs (with or without the actual uml-kernel, as you choose), and
something similar to the toms-uml.tar.gz I just sent out; the build script
and the necessary changes to build a root_fs.

	As a side note, what would you think of splitting the kernel from
the root_fs instead of making them a unified package?  That would save you
the effort of rebuilding the unified package each time a new kernel comes
out.  On the other hand, it means an extra step for the users.  6 of one,
half dozen of the other...
	Cheers,
	- Bill

---------------------------------------------------------------------------
/*
 * Leonard, I'll tie you, draw around you a pentagram
 * and read this file. Aloud.
 */
	- from kernel 2.3.40-pre6's drivers/block/DAC960.c
--------------------------------------------------------------------------
William Stearns (wst...@po...).  Mason, Buildkernel, named2hosts, 
and ipfwadm2ipchains are at:                http://www.pobox.com/~wstearns
LinuxMonth; articles for Linux Enthusiasts! http://www.linuxmonth.com
--------------------------------------------------------------------------

Re: [User-mode-linux-devel] New root_fs; problems with Toms Root/boot (libc5)

[Jeff D. <jd...@ka...>]

I got it to boot with no changes to the disk itself except for adding an 
/etc/passwd (not even /dev/ubd/*, which I forgot :-).

Your immediate problem is probably that I have the old select syscall hooked 
up wrong.  It goes into an infinite loop returning an error because the 
arguments are being picked up wrong in the kernel.  Fixing that and another 
problem gives me a clean boot.

If you are building from sources, I can give you patches which fix things.

And what's with no /etc/passwd, anyway?  Is there some other way to get a 
shell out of it?

> I'd love to see it as one of the available root environments for uml.
> He does have a short copyright message and two urls that he'd like
> included in any derivative works, but I suspect that's not too much of
> a problem; they could be included in a short README.toms.

OK, I can stick it up there once I put up a kernel that can boot it.

				Jeff

Re: [User-mode-linux-devel] New root_fs; problems with Toms Root/boot (libc5)

[William S. <wst...@po...>]

Good morning, Jeff,

On Sun, 18 Jun 2000, Jeff Dike wrote:

> > Try booting with 'devfs=nomount' and see if it  works better.

	OK, that helped.

> I forgot to mention that you'll also need to add a ubd device or two (block 
> major 98) to /dev.

	That didn't seem to make a difference to my current problem;
/sbin/init seems to execute rc.S but not get any further.  I added an
"echo Exiting rc.S" to the end and that shows up on the screen, but the
corresponding "entering rc.M" does not show up on the screen.  Init and
the tracing thread start fighting for CPU time.
	Here's a snippet from strace -p 6081 (the tracing thread).  6083
is init.  I'm not sure whether this is normal or abnormal:

--- SIGCHLD (Child exited) ---
wait4(-1, [WIFSTOPPED(s) && WSTOPSIG(s) == SIGTRAP], WUNTRACED, NULL) =
6083
gettimeofday({961299583, 134497}, NULL) = 0
ptrace(PTRACE_PEEKUSER, 6083, 4*EIP, [0x4005b1b4]) = 0
ptrace(PTRACE_???, 6083, 0, 0x50fe827c) = 0
ptrace(PTRACE_???, 6083, 0, 0x50fe82c0) = 0
ptrace(PTRACE_POKEUSER, 6083, 4*ORIG_EAX, 0x14) = 0
ptrace(PTRACE_CONT, 6083, 0, SIG_0)     = 0
--- SIGCHLD (Child exited) ---
wait4(-1, [WIFSTOPPED(s) && WSTOPSIG(s) == SIGUSR1], WUNTRACED, NULL) =
6083
gettimeofday({961299583, 139633}, NULL) = 0
ptrace(PTRACE_PEEKUSER, 6083, 4*EIP, [0x100a4541]) = 0
ptrace(PTRACE_???, 6083, 0, 0x50fe827c) = 0
ptrace(PTRACE_SYSCALL, 6083, 0, SIG_0)  = 0
--- SIGCHLD (Child exited) ---
wait4(-1, [WIFSTOPPED(s) && WSTOPSIG(s) == SIGTRAP], WUNTRACED, NULL) =
6083
gettimeofday({961299583, 142397}, NULL) = 0
ptrace(PTRACE_PEEKUSER, 6083, 4*EIP, [0x4005b1b4]) = 0
ptrace(PTRACE_???, 6083, 0, 0x50fe827c) = 0
ptrace(PTRACE_???, 6083, 0, 0x50fe82c0) = 0
ptrace(PTRACE_POKEUSER, 6083, 4*ORIG_EAX, 0x14) = 0
ptrace(PTRACE_CONT, 6083, 0, SIG_0)     = 0
--- SIGCHLD (Child exited) ---
(repeat, ad infinitum)

	By adding "init=/bin/ash" to the command line, I can get to a
command prompt.  Hey, we're getting somewhere, bit by bit.

> BTW, what is Tom's Root/boot?  Is there anything special about it?  Would it 
> make sense to add it to my downloads when it can boot uml?

	Toms Root/boot (available at ftp.toms.net) is a single floppy
Linux.  Tom Oehser <to...@to...> has done a marvelous job of packing an
entire recovery environment into a 1.7M floppy.  It's really indespensible
for emergencies.
	Ny using a libc5 environment and doing insane things to shrink the
binaries, he's succeded at getting a very functional Linux system on a
single floppy.

	I'd love to see it as one of the available root environments for
uml.  He does have a short copyright message and two urls that he'd like
included in any derivative works, but I suspect that's not too much of a
problem; they could be included in a short README.toms.
	Quite honestly, if the last few details can be worked out, the
availability of a 3.6M root filesystem is a nice plus for simple projects.  
While I'm all for full distributions and their features, a slim root_fs
should be useful in itself.

	I've attached a tar file of the current work.  I'll be traveling
for a few days, so if anybody feels like playing, feel free.  I'd love to
see what people find...

mkdir --parents /mnt/spare/mirrors/	#Or change workpath in the script
cd /mnt/spare/mirrors
tar -xzvf toms-uml.tar.gz
cd /mnt/spare/mirrors/toms-uml

	If you don't have wget on your system, you'll have to
manually download tomsrtbt-1.7.205.tar.gz from ftp.toms.net to
/mnt/spare/mirrors/toms-uml
	Copy/link the uml kernel into /mnt/spare/mirrors/toms-uml .

./mk_toms_root 1.7.205

	Cheers,
	- Bill

---------------------------------------------------------------------------
	I called up the Bureau of Alcohol, Tobacco, and Firearms regional
office and asked, "What wine goes best with an M-16?" The guy who
answered did his best to be helpful: "That depends. What are you smoking?"
(Courtesy of Andrej Todosic <ato...@ub...>)
--------------------------------------------------------------------------
William Stearns (wst...@po...).  Mason, Buildkernel, named2hosts, 
and ipfwadm2ipchains are at:                http://www.pobox.com/~wstearns
LinuxMonth; articles for Linux Enthusiasts! http://www.linuxmonth.com
--------------------------------------------------------------------------

Re: [User-mode-linux-devel] New root_fs; problems with Toms Root/boot (libc5)

[Jeff D. <jd...@ka...>]

> Try booting with 'devfs=nomount' and see if it  works better.

I forgot to mention that you'll also need to add a ubd device or two (block 
major 98) to /dev.

BTW, what is Tom's Root/boot?  Is there anything special about it?  Would it 
make sense to add it to my downloads when it can boot uml?

				Jeff

Re: [User-mode-linux-devel] New root_fs; problems with Toms Root/boot (libc5)

[Jeff D. <jd...@ka...>]

wst...@po... said:
> unable to allocate pty/tty pair

Do you have the kernel mounting devfs?  If so, that will cover up /dev, and 
things likely won't work.  Try booting with 'devfs=nomount' and see if it 
works better.

				Jeff

Re: [User-mode-linux-devel] New root_fs; problems with Toms Root/boot (libc5)

[William S. <wst...@po...>]

Good evening, Jeff,

On Sat, 17 Jun 2000, William Stearns wrote:

> 	I'm trying to set up Tom's root/boot root filesystem as a uml
> root_fs.  I'm getting an error during the boot process; it looks like
> something in init, but I don't know what.  The kernel is 2.3.99-pre9-1um.
> 
> INIT: version 2.60 booting
> INIT: Switching to runlevel: 5
> /usr/doc/*.gz: No such file or directory
> .: Can't open /usr/doc/rc.custom
> Kernel panic: Kernel mode fault at addr 0x0, ip 0xbf7ffcec
> Kernel panic: fix_range fixing wrong address space

	This comes from the end of /etc/rc.M .  This is the only modified
file, so far, from the original distribution; because Tomsrtbt runs
completely off a 1.7M floppy, it has to go through some conniptions to
decompress the filesystem.  I comment out the uncompress lines.  The last
three lines:

gzip -d /usr/doc/*.gz
. /usr/doc/rc.custom
#umount /fl

	OK, so rc.M finishes and control transfers back to init.  Here's
the entire inittab:

id:4:initdefault:
si:S:sysinit:/etc/rc.S
rc:5:wait:/etc/rc.M
ca::ctrlaltdel:/bin/shutdown -t5 -rfn now "CtlAltDel"
l0:0:wait:/etc/rc.0
l6:6:wait:/etc/rc.6
c1:5:respawn:/usr/bin/agetty 38400 tty1
c2:5:respawn:/usr/bin/agetty 38400 tty2
c3:5:respawn:/usr/bin/agetty 38400 tty3
c4:5:respawn:/usr/bin/agetty 38400 tty4

	Is agetty the problem?  AHA!  Change the last 4 lines to:

c1:5:respawn:/usr/bin/agetty 38400 console
#c2:5:respawn:/usr/bin/agetty 38400 tty2
#c3:5:respawn:/usr/bin/agetty 38400 tty3
#c4:5:respawn:/usr/bin/agetty 38400 tty4

	and we get:

/usr/doc/*.gz: No such file or directory
.: Can't open /usr/doc/rc.custom
(none) login: root
/bin/login: cannot open /etc/passwd: no such file
LOGIN FAILED

	OK, so uml doesn't like tty1-4?

	Back to LOGIN FAILED; add a password file with an empty root
password and:

/usr/doc/*.gz: No such file or directory
.: Can't open /usr/doc/rc.custom
(none) login: root
Password: ile rev.2.01
ile: unable to allocate pty/tty pair
(none) login: 

	Hmmm.  Why?  Here are the pty/tty's from the dev directory:

[root@sparrow dev]# ls -al [pt]ty* | less
crw-rw-rw-    1 root     root       2,   0 May  7  1999 ptyp0
crw-rw-rw-    1 root     root       2,   1 May  7  1999 ptyp1
crw-rw-rw-    1 root     root       2,   2 May  7  1999 ptyp2
crw-rw-rw-    1 root     root       2,   3 May  7  1999 ptyp3
crw-rw-rw-    1 root     root       2,   4 May  7  1999 ptyp4
crw-rw-rw-    1 root     root       2,   5 May  7  1999 ptyp5
crw-rw-rw-    1 root     root       2,   6 May  7  1999 ptyp6
crw-rw-rw-    1 root     root       2,   7 May  7  1999 ptyp7
crw-rw-rw-    1 root     root       5,   0 May  7  1999 tty
crw-rw-rw-    1 root     root       4,   0 May  8  1999 tty0
crw-rw-rw-    1 root     root       4,   1 May  7  1999 tty1
crw-rw-rw-    1 root     root       4,   2 May  7  1999 tty2
crw-rw-rw-    1 root     root       4,   3 May  7  1999 tty3
crw-rw-rw-    1 root     root       4,   4 May  7  1999 tty4
crw-rw-rw-    1 root     root       4,   5 May  7  1999 tty5
crw-rw-rw-    1 root     root       4,   6 May  7  1999 tty6
crw-rw-rw-    1 root     root       4,   7 May  7  1999 tty7
crw-rw-rw-    1 root     root       4,   8 May  7  1999 tty8
crw-rw-r--    1 root     root       4,  64 Mar  1  1998 ttyS0
crw-rw-rw-    1 root     root       4,  65 Apr 28  1999 ttyS1
crw-r--r--    1 root     root       4,  66 May  7  1999 ttyS2
crw-rw-r--    1 root     root       4,  67 May  7  1999 ttyS3
crw-rw-rw-    1 root     root       3,   0 May  7  1999 ttyp0
crw-rw-rw-    1 root     root       3,   1 May  7  1999 ttyp1
crw-rw-rw-    1 root     root       3,   2 May  7  1999 ttyp2
crw-rw-rw-    1 root     root       3,   3 May  7  1999 ttyp3
crw-rw-rw-    1 root     root       3,   4 May  7  1999 ttyp4
crw-rw-rw-    1 root     root       3,   5 May  7  1999 ttyp5
crw-rw-rw-    1 root     root       3,   6 May  7  1999 ttyp6
crw-rw-rw-    1 root     root       3,   7 May  7  1999 ttyp7

	And now I'm lost.  The only changes I've made from the file I
mailed you are the passwd file:

root::0:0:root:/root:/bin/ash
guest::500:500:guest:/tmp:/bin/ash

	and the 4 line change at the end of inittab I listed above.
	Cheers,
	- Bill

---------------------------------------------------------------------------
	Like the ad says, at 300 dpi you can tell she's wearing a
swimsuit. At 600 dpi you can tell it's wet. At 1200 dpi you can tell it's
painted on. I suppose at 2400 dpi you can tell if the paint is giving her
a rash. (So says Joshua R. Poulson) 
(Courtesy of Bob Taylor <brt...@qt...>)
--------------------------------------------------------------------------
William Stearns (wst...@po...).  Mason, Buildkernel, named2hosts, 
and ipfwadm2ipchains are at:                http://www.pobox.com/~wstearns
LinuxMonth; articles for Linux Enthusiasts! http://www.linuxmonth.com
--------------------------------------------------------------------------

[User-mode-linux-devel] Re: New root_fs; problems with Toms Root/boot (libc5)

[William S. <wst...@po...>]

On Sat, 17 Jun 2000, Jeff Dike wrote:

> wst...@po... said:
> > Kernel panic: Kernel mode fault at addr 0x0, ip 0xbf7ffcec 
> 
> I'm on a search-and-destroy mission for that bug (and a couple of others) 
> right now.  How consistently can you reproduce it?

	100%, every time it comes up.  It never goes further.
	bzip2 file on its way under seperate cover...
	Cheers,
	- Bill

---------------------------------------------------------------------------
	I'm not tense, just terribly, terribly alert.
(Courtesy of "Michael J. Dark" <dar...@ze...>)
--------------------------------------------------------------------------
William Stearns (wst...@po...).  Mason, Buildkernel, named2hosts, 
and ipfwadm2ipchains are at:                http://www.pobox.com/~wstearns
LinuxMonth; articles for Linux Enthusiasts! http://www.linuxmonth.com
--------------------------------------------------------------------------

[User-mode-linux-devel] Re: New root_fs; problems with Toms Root/boot (libc5)

[Jeff D. <jd...@ka...>]

wst...@po... said:
> Kernel panic: Kernel mode fault at addr 0x0, ip 0xbf7ffcec 

I'm on a search-and-destroy mission for that bug (and a couple of others) 
right now.  How consistently can you reproduce it?

				Jeff

[User-mode-linux-devel] New root_fs; problems with Toms Root/boot (libc5)

[William S. <wst...@po...>]

Good day, Jeff,
	I'm trying to set up Tom's root/boot root filesystem as a uml
root_fs.  I'm getting an error during the boot process; it looks like
something in init, but I don't know what.  The kernel is 2.3.99-pre9-1um.

INIT: version 2.60 booting
INIT: Switching to runlevel: 5
/usr/doc/*.gz: No such file or directory
.: Can't open /usr/doc/rc.custom
Kernel panic: Kernel mode fault at addr 0x0, ip 0xbf7ffcec
Kernel panic: fix_range fixing wrong address space

	The first comes from uml's trap_kern.c, the second from tlb.c.  
I'd be glad to mail you the 1.7M root_fs.bz2 if you're interested.
	Any ideas?
	Cheers,
	- Bill

---------------------------------------------------------------------------
	"Windows 95 will now attempt to blow chunks across your primary
partition.  Press any key to continue..." 
(Courtesy of "Eric Princen" <epr...@ma...>)
--------------------------------------------------------------------------
William Stearns (wst...@po...).  Mason, Buildkernel, named2hosts, 
and ipfwadm2ipchains are at:                http://www.pobox.com/~wstearns
LinuxMonth; articles for Linux Enthusiasts! http://www.linuxmonth.com
--------------------------------------------------------------------------