Godmar Back [Fri, 16 Nov 2007 02:18:35 +0000 (02:18 +0000)]
added IGNORE_USER_FAULTS clause to match specification
Godmar Back [Sun, 23 Sep 2007 15:40:23 +0000 (15:40 +0000)]
fixed typo: "pass" should be "fail"
Ben Pfaff [Fri, 18 May 2007 23:23:51 +0000 (23:23 +0000)]
Clean some more files.
Ben Pfaff [Mon, 7 May 2007 23:02:07 +0000 (23:02 +0000)]
Fix warning.
Ben Pfaff [Mon, 23 Apr 2007 00:05:33 +0000 (00:05 +0000)]
Correctly initialize the esp0 pointer in the TSS for the initial
kernel thread. Shouldn't actually matter in practice, because that
pointer is only used on a user->kernel transition, and the initial
kernel thread never runs a user process.
Reported by Godmar Back.
Ben Pfaff [Mon, 2 Apr 2007 08:10:25 +0000 (08:10 +0000)]
Ignore tags and cscope files.
Ben Pfaff [Mon, 2 Apr 2007 07:23:22 +0000 (07:23 +0000)]
Improve comment on thread_unblock().
Suggested by Godmar Back.
Ben Pfaff [Sat, 31 Mar 2007 14:40:52 +0000 (14:40 +0000)]
Fix bugs in -j and -r options reported by Godmar Back.
Ben Pfaff [Sat, 31 Mar 2007 14:40:29 +0000 (14:40 +0000)]
Fix bug in jitter reported by Godmar Back (BIOS wouldn't finish
initializing).
Ben Pfaff [Sat, 31 Mar 2007 14:12:03 +0000 (14:12 +0000)]
Fix specification of "write" system call to match the file system
implementation and the tests.
Ben Pfaff [Sat, 31 Mar 2007 13:49:21 +0000 (13:49 +0000)]
Describe cscope.
Ben Pfaff [Sat, 31 Mar 2007 12:26:45 +0000 (12:26 +0000)]
Delete question about pwd.c program, as agreed with Godmar earlier.
Ben Pfaff [Sat, 24 Mar 2007 01:26:18 +0000 (01:26 +0000)]
Improve wording.
From Godmar Back.
Ben Pfaff [Sat, 24 Mar 2007 01:12:10 +0000 (01:12 +0000)]
Explain the semantics of exec in more detail.
From Godmar Back.
Ben Pfaff [Sat, 24 Mar 2007 01:04:19 +0000 (01:04 +0000)]
Add helpful comment.
From Godmar Back.
Ben Pfaff [Sat, 24 Mar 2007 01:02:19 +0000 (01:02 +0000)]
Add "cscope" target.
Ben Pfaff [Sat, 24 Mar 2007 01:01:06 +0000 (01:01 +0000)]
Rename execute_thread() to start_process().
Thanks to Godmar Back for pointing out that it could be better named.
Ben Pfaff [Sat, 24 Mar 2007 00:56:57 +0000 (00:56 +0000)]
Clarify that create and remove work on files, not on file descriptors.
Thanks to Godmar Back for pointing out the need.
Ben Pfaff [Thu, 8 Mar 2007 03:56:24 +0000 (03:56 +0000)]
Fix bug in *-persistence tests when prerequisites fail.
Reported by Tanvi Vyas <tanvi@stanford.edu>.
Ben Pfaff [Wed, 7 Mar 2007 14:52:06 +0000 (14:52 +0000)]
Fix inconstiency in recent_cpu update description.
Ben Pfaff [Sun, 11 Feb 2007 04:45:07 +0000 (04:45 +0000)]
Fix typo.
Ben Pfaff [Sun, 11 Feb 2007 04:44:25 +0000 (04:44 +0000)]
Clarify description of "wait" syscall. Thanks to Evan Kutter
<ejk@stanford.edu> for describing the issue.
Ben Pfaff [Tue, 6 Feb 2007 21:09:10 +0000 (21:09 +0000)]
Improve example for "Can a thread's priority change while it is on the
ready queue?" FAQ. From Godmar Back.
Godmar Back [Tue, 6 Feb 2007 06:44:01 +0000 (06:44 +0000)]
added patch to make bochs work with gdb 6.5
Ben Pfaff [Sun, 28 Jan 2007 04:21:55 +0000 (04:21 +0000)]
Fix order of checks on load average.
From a CS140 student, via Ben Sapp.
Ben Pfaff [Sun, 21 Jan 2007 10:29:26 +0000 (10:29 +0000)]
Add i86pc to unames detected as x86 (used by Solaris).
From Michael Gregory <mpgreg@stanford.edu>.
Godmar Back [Sat, 13 Jan 2007 14:39:41 +0000 (14:39 +0000)]
fixed grammar/typo
Ben Pfaff [Thu, 11 Jan 2007 08:02:13 +0000 (08:02 +0000)]
Set VERBOSE to empty string by default, which prevents an environment
variable named VERBOSE from causing trouble but still allows it to be
overridden by the user on the command line. (Some of the Stanford
machines set VERBOSE=yes in the environment by default on log-in.)
Ben Pfaff [Thu, 11 Jan 2007 07:48:07 +0000 (07:48 +0000)]
Remove unnecessary semicolon that gives GNU make 3.80 fits.
Ben Pfaff [Thu, 11 Jan 2007 06:57:40 +0000 (06:57 +0000)]
Don't call bx_gdbstub_exception when we're not compiling the gdb stub.
Thanks to Godmar Back for pointing this out.
Ben Pfaff [Tue, 9 Jan 2007 17:46:42 +0000 (17:46 +0000)]
Add -MF to GCC invocation to make ccache happy on Fedora Core 6.
Godmar Back [Mon, 8 Jan 2007 09:55:17 +0000 (09:55 +0000)]
clarified stack limit issue and removed comment about first instruction faulting in stack page
Godmar Back [Mon, 8 Jan 2007 08:46:49 +0000 (08:46 +0000)]
- removed @ref to "printf Reboots" since this no longer applies and the @anchor is gone
- fixed typo print -> printf
Godmar Back [Mon, 8 Jan 2007 06:11:37 +0000 (06:11 +0000)]
adjusted scores for P4 as discussed
Godmar Back [Sun, 7 Jan 2007 21:38:02 +0000 (21:38 +0000)]
Cosmetic fix: make first column wider to account for longer names in p4.
Godmar Back [Sun, 7 Jan 2007 20:55:28 +0000 (20:55 +0000)]
- adjustments to scores for P3 as discussed.
- increased problem size for page-* tests to force eviction
Godmar Back [Sun, 7 Jan 2007 20:54:54 +0000 (20:54 +0000)]
adjustments to scores for P2 as discussed.
Godmar Back [Sat, 6 Jan 2007 19:00:52 +0000 (19:00 +0000)]
- added priority-donate-chain test which tests 8-level deep nested donation
- adjusted weights for P1 Grading file
Ben Pfaff [Thu, 4 Jan 2007 15:13:49 +0000 (15:13 +0000)]
Separate persistence tests from the extended filesys tests.
Suggested by Godmar Back.
Ben Pfaff [Thu, 4 Jan 2007 14:19:25 +0000 (14:19 +0000)]
New page-merge-mm, page-merge-stk tests.
Thanks to Godmar Back for suggestion.
Ben Pfaff [Mon, 18 Dec 2006 18:33:52 +0000 (18:33 +0000)]
Make comment more readable.
Ben Pfaff [Fri, 15 Dec 2006 19:40:03 +0000 (19:40 +0000)]
Fix expected output.
Ben Pfaff [Fri, 15 Dec 2006 14:58:10 +0000 (14:58 +0000)]
Don't require distinguishing data and metadata for caching.
Another suggestion for implementation order.
From Godmar.
Ben Pfaff [Fri, 15 Dec 2006 14:24:48 +0000 (14:24 +0000)]
Fix comment.
Ben Pfaff [Fri, 15 Dec 2006 14:21:04 +0000 (14:21 +0000)]
When printing a directory tree, indicate explicitly if it's empty.
Ben Pfaff [Mon, 11 Dec 2006 20:40:20 +0000 (20:40 +0000)]
Parse command-line options much earlier, so that thread_mlfqs is
initialized before thread_init() gets hold of it.
Also allow console output much earlier, by doing initialization
whenever we call into it first and by only trying to take the console
lock after threads have been initialized.
Don't try to take the console lock after the kernel panics, to avoid
getting nailed so hard by bugs in lock_acquire().
Suggested by Godmar Back.
Ben Pfaff [Fri, 8 Dec 2006 23:40:23 +0000 (23:40 +0000)]
New target to generate cscope.files.
Ben Pfaff [Fri, 8 Dec 2006 18:47:37 +0000 (18:47 +0000)]
Make it possible to ignore page faults in GDB.
From su.class.cs140 and Godmar Back.
Ben Pfaff [Fri, 8 Dec 2006 16:06:31 +0000 (16:06 +0000)]
Revert the bitmap randomization change, which had the side effect of
causing enough disk fragmentation for the base (extent-based) file
system that in some cases tests would fail. Also, it confused the
hell out of everyone. Failed experiment.
Ben Pfaff [Fri, 8 Dec 2006 15:39:32 +0000 (15:39 +0000)]
When reading files out of the scratch disk, don't die with an error
exit code if there's a bad signature, because that's a guest error not
a host error. (If we do exit with an error code, it fouls up the
grading process.) Instead, just make sure that the host file(s) we
were supposed to retrieve is unlinked.
Ben Pfaff [Fri, 8 Dec 2006 15:29:58 +0000 (15:29 +0000)]
Fix error message when archive fails to open.
Ben Pfaff [Fri, 8 Dec 2006 14:58:02 +0000 (14:58 +0000)]
Explain that multiple file descriptors for a single file are
independent.
From Godmar Back.
Ben Pfaff [Fri, 8 Dec 2006 14:52:37 +0000 (14:52 +0000)]
Note that struct file, struct dir are per-thread.
From Godmar Back.
Ben Pfaff [Fri, 8 Dec 2006 14:48:20 +0000 (14:48 +0000)]
Add Project 4 Suggested Order of Implementation section,
from slides written by Godmar Back.
Ben Pfaff [Fri, 8 Dec 2006 14:35:09 +0000 (14:35 +0000)]
Explain better what we expect from the buffer cache.
Ben Pfaff [Mon, 4 Dec 2006 19:14:26 +0000 (19:14 +0000)]
Use mkdir, not create.
Thanks to "Yuriy" for reporting this bug.
Ben Pfaff [Tue, 21 Nov 2006 13:04:33 +0000 (13:04 +0000)]
Clarify condition on BUF_SIZE in monitor example.
Thanks to Terje Mathiesen for pointing out this issue.
Ben Pfaff [Sat, 11 Nov 2006 14:44:42 +0000 (14:44 +0000)]
Add comment.
Ben Pfaff [Sat, 11 Nov 2006 14:43:15 +0000 (14:43 +0000)]
Drop use of volatile in favor of explicit memory barriers.
Document this use of barriers.
Ben Pfaff [Sat, 11 Nov 2006 14:41:59 +0000 (14:41 +0000)]
Also verify that running "make clean" deletes all the files in the
build directory except for the Makefile itself.
Ben Pfaff [Sat, 11 Nov 2006 14:30:18 +0000 (14:30 +0000)]
Clean tests/vm/zeros.
Ben Pfaff [Sat, 11 Nov 2006 14:29:14 +0000 (14:29 +0000)]
Adopt consistent style in asm statements.
Ben Pfaff [Sat, 11 Nov 2006 14:25:16 +0000 (14:25 +0000)]
Add memory clobbers to several asm statements,
to prevent reordering.
Ben Pfaff [Sat, 11 Nov 2006 14:19:40 +0000 (14:19 +0000)]
Make sure loops_per_tick doesn't overflow to zero.
Ben Pfaff [Fri, 10 Nov 2006 16:53:21 +0000 (16:53 +0000)]
Also clean results, grade.
Ben Pfaff [Fri, 10 Nov 2006 15:51:03 +0000 (15:51 +0000)]
Add "memory" as clobber to the asm used by barrier().
Otherwise at least some versions of GCC assume that no changes to
memory can take place.
Ben Pfaff [Fri, 10 Nov 2006 15:31:18 +0000 (15:31 +0000)]
Also clean bochsout.txt, bochsrc.txt.
Ben Pfaff [Fri, 10 Nov 2006 15:06:15 +0000 (15:06 +0000)]
Minor style updates.
Godmar Back [Fri, 10 Nov 2006 04:36:14 +0000 (04:36 +0000)]
Rewrite of multi-oom. Now descends 10 times and expects to reach the same
depth every time. In addition, fires off some children that consume resources
and then crash. Much more difficult than before.
Godmar Back [Fri, 10 Nov 2006 04:34:20 +0000 (04:34 +0000)]
- when ignoring exit codes, ignore exit(-1) also (previously didn't allow '-')
Godmar Back [Fri, 10 Nov 2006 02:20:08 +0000 (02:20 +0000)]
make loop counter volatile to prevent gcc4.1 from optimizing away entire loop body
Godmar Back [Fri, 10 Nov 2006 02:18:07 +0000 (02:18 +0000)]
added accidentally omitted $ to make operand immediate
Godmar Back [Sat, 4 Nov 2006 20:08:27 +0000 (20:08 +0000)]
clarify use of timer_sleep or acceptable alternatives for implementing
write-behind cache flushing.
Ben Pfaff [Fri, 3 Nov 2006 03:07:56 +0000 (03:07 +0000)]
Clarify purpose and rationale.
From Godmar Back.
Ben Pfaff [Fri, 3 Nov 2006 02:54:48 +0000 (02:54 +0000)]
Remove O_EXCL from flags used to create files to be copied out of a
Pintos scratch disk, in get_scratch_file.
Thanks to Godmar Back for pointing out the problem.
Ben Pfaff [Fri, 3 Nov 2006 02:50:36 +0000 (02:50 +0000)]
Delete .tar, .get-output, and .get-errors files on "clean" target.
Thanks to Godmar Back for pointing out the problem.
Godmar Back [Wed, 1 Nov 2006 21:20:52 +0000 (21:20 +0000)]
added IGNORE_USER_FAULTS flag - so students may report failed stack growth
Godmar Back [Wed, 1 Nov 2006 00:00:42 +0000 (00:00 +0000)]
corrected wrong statement about content of f->esp when page fault
occurs in kernel mode. See [IA32-v3a], figure 5.4 in section 5.12.1.1.
Ben Pfaff [Sat, 7 Oct 2006 19:01:09 +0000 (19:01 +0000)]
Improve comment.
Ben Pfaff [Thu, 5 Oct 2006 03:07:31 +0000 (03:07 +0000)]
Sleep until recent_cpu actually decays.
Ben Pfaff [Thu, 5 Oct 2006 03:05:40 +0000 (03:05 +0000)]
Add hints about not doing too much work in timer interrupt.
Ben Pfaff [Thu, 5 Oct 2006 02:20:10 +0000 (02:20 +0000)]
Fix message.
From Godmar Back.
Ben Pfaff [Thu, 5 Oct 2006 01:25:38 +0000 (01:25 +0000)]
When out of memory+swap, just panic the kernel.
Ben Pfaff [Thu, 5 Oct 2006 01:14:14 +0000 (01:14 +0000)]
Add -O to CFLAGS to try to catch uninitialized variable errors.
Ben Pfaff [Thu, 5 Oct 2006 00:44:09 +0000 (00:44 +0000)]
Make Pintos able to build with "gcc -m32" on x86-64 hosts without the
need for the host to have any 32-bit libraries (specifically, without
needing libgcc).
Ben Pfaff [Tue, 3 Oct 2006 09:41:00 +0000 (09:41 +0000)]
Fix logarithms.
Thanks to Haobo Xu <haoboxu@stanford.edu>.
Ben Pfaff [Thu, 28 Sep 2006 17:45:27 +0000 (17:45 +0000)]
Irq 7 is special. It is not just the parallel port. It can also be a
"spurious" irq, one that is not associated with any particular device.
Real hardware can generate spurious irqs for various reasons (try
Google for details), but it never occurred to me that simulated
hardware would ever do so.
However, it can, at least under Bochs. Pintos programs the PIT to
interrupt at 100 Hz. The PIT does so by generating a output signal
that stays high most of the time but drops low for a single clock
cycle 100 times per second. The low-to-high transition causes the
timer interrupt.
The problem occurs when we've been doing work with interrupts
disabled, causing a timer interrupt to be delayed. If we enable
interrupts on the same clock cycle as the PIT's high-to-low
transition, we will still get an interrupt, because the interrupt
controller latched the presence of an interrupt. However, when the
CPU attempts to determine the irq number, the interrupt controller
sees that all of its incoming interrupt lines are low. Thus, it
reports irq 7 to the CPU, and we end up with an interrupt 0x27.
It doesn't happen very often, because there's only a 1-cycle window,
and it requires that we do a good deal of work with interrupts off,
and it can only happen if no other interrupts (e.g. serial) are
pending. But it's darned annoying when it happens.
Here's a filtered excerpt from a Bochs log with debugging enabled
where you can see the problem. The first interrupt (32 or 0x20) is
properly a timer interrupt, because irq line 0 is high. The second
interrupt (39 or 0x27) is the spurious one; you can see that irq line
0 went low on the same cycle.
00080033515d[PIC ] IRQ line 4 now low
00080033827d[PIC ] IRQ line 4 now high
00080033970d[PIC ] IRQ line 0 now low
00080033971d[PIC ] IRQ line 0 now high
00080033989d[PIC ] IRQ line 4 now low
00080034068d[CPU ] interrupt(): vector = 32, INT = 0, EXT = 1
00080043970d[PIC ] IRQ line 0 now low
00080043971d[PIC ] IRQ line 0 now high
00080053971d[PIC ] IRQ line 0 now low
00080053971d[PIC ] IRQ line 0 now high
00080063971d[PIC ] IRQ line 0 now low
00080063972d[PIC ] IRQ line 0 now high
00080073971d[PIC ] IRQ line 0 now low
00080073972d[PIC ] IRQ line 0 now high
00080083971d[PIC ] IRQ line 0 now low
00080083971d[CPU ] interrupt(): vector = 39, INT = 0, EXT = 1
00080083972d[PIC ] IRQ line 0 now high
00080084538d[PIC ] IRQ line 4 now high
This change causes irqs 7 and 15 (0x27 and 0x2f) to be ignored when no
handler is provided.
Problem reported by Godmar Back.
Ben Pfaff [Thu, 28 Sep 2006 13:11:52 +0000 (13:11 +0000)]
Update.
Ben Pfaff [Thu, 28 Sep 2006 12:59:05 +0000 (12:59 +0000)]
Minor doc improvements.
From Godmar Back.
Ben Pfaff [Thu, 28 Sep 2006 12:36:41 +0000 (12:36 +0000)]
Make list_entry, hash_entry more type-safe.
Suggestion from Godmar Back.
Ben Pfaff [Thu, 28 Sep 2006 12:15:11 +0000 (12:15 +0000)]
Fix confusing spacing in example
Ben Pfaff [Tue, 26 Sep 2006 15:12:05 +0000 (15:12 +0000)]
Create idle thread with minimum priority.
Suggested by Godmar Back.
Ben Pfaff [Tue, 26 Sep 2006 14:59:41 +0000 (14:59 +0000)]
Change "Memory Barriers" to "Optimization Barriers".
Thanks to Joshua Haberman for pointing out the issue.
Ben Pfaff [Mon, 25 Sep 2006 21:43:09 +0000 (21:43 +0000)]
Move assignment to _end_kernel_text inside .rodata.
Without this change, any orphan, read-only sections (such as .eh_frame
from libgcc) will be placed by GNU ld between .rodata and
_end_kernel_text. The result is that _end_kernel_text gets pushed
into the middle of a page that now contains some read-only and some
read/write data. After paging_init() marks that page read-only, the
first write to it (in intr_init() in my case) causes Pintos to go
ka-boom!
With this change, orphan read-only sections just don't get marked
read-only in the page tables. Big deal.
Ben Pfaff [Mon, 25 Sep 2006 20:26:35 +0000 (20:26 +0000)]
Fix two bugs in the base Pintos code:
1. Idle thread doesn't get initialized well: it hangs around on the
ready list until we have idle time. Fix by using a semaphore to
make sure the idle thread gets initialized.
2. After idle thread does get initialized, it can get put back on the
ready list if you yield from the timer interrupt. Fix by not ever
putting the idle thread on the ready list.
Plus fix a bug in the sample solution that tends to mask #2:
3. Doesn't yield from timer interrupt when unblocking a thread. Fix
by yielding from timer interrupt when unblocking a thread (of
higher priority).
Also, fix something that confused students:
4. Rename enable_mlfqs to thread_mlfqs and move it to thread.c (to
reduce confusion about where it is declared), and heavily comment
to ensure that students know when it gets initialized.
These problems were drawn to my attention by Godmar Back. Thanks!
Godmar Back [Wed, 13 Sep 2006 11:40:40 +0000 (11:40 +0000)]
fixed left-over, now-wrong comment about how decrease in numeric priority means increase in effective priority
Godmar Back [Tue, 12 Sep 2006 02:32:27 +0000 (02:32 +0000)]
fixed typo in wrong iterator example
Ben Pfaff [Sat, 9 Sep 2006 16:53:56 +0000 (16:53 +0000)]
Fix issue reported by Godmar Back:
I may have found a problem with the triple fault patch, though I don't
know how to solve it.
If you add a *(int *)0=42 right after paging_init(), but before
intr_init(), gdb will just hang:
...
where it hangs while Pintos prints:
Triple fault: stopping for gdb
Triple fault: stopping for gdb
Triple fault: stopping for gdb
Triple fault: stopping for gdb
Triple fault: stopping for gdb
Triple fault: stopping for gdb
Triple fault: stopping for gdb
Triple fault: stopping for gdb
Triple fault: stopping for gdb
Triple fault: stopping for gdb
Triple fault: stopping for gdb
etc.
It stops properly if I don't next-step (that is, if I just let it run.)
Ben Pfaff [Sat, 9 Sep 2006 16:52:39 +0000 (16:52 +0000)]
Convert tabs to spaces to make cut-and-paste to interactive shell
work.
Ben Pfaff [Fri, 8 Sep 2006 00:51:17 +0000 (00:51 +0000)]
Make kernel code pages read-only.
From Godmar Back, with changes.