*ERROR* Cannot initialize the agpgart module.

Dave when you have this in a bk tree let me know and I will pull and verify it working here.

Thanks to both of you

Ed Tomlinson

bk://linux-dj.bkbits.net/agpgart

I've given it a compile testing, but not booted it yet. Scream if necessary.

Dave

| Dave Jones.        http://www.codemonkey.org.uk

Dave, with the pull from this morning (8am EST), it almost works modular. I get:

Dec 20 18:20:19 oscar upsd[636]: Communication established
Dec 20 18:20:47 oscar kernel: Linux agpgart interface v0.100 (c) Dave Jones
Dec 20 18:20:47 oscar kernel: agpgart: Detected VIA MVP3 chipset
Dec 20 18:20:47 oscar kernel: agpgart: AGP aperture is 64M @ 0xe0000000
Dec 20 18:20:58 oscar kernel: [drm] Initialized mga 3.1.0 20021029 on minor 0
Dec 20 18:20:58 oscar kernel: Module agpgart cannot be unloaded due to unsafe usage in drivers/char/ag

p/backend.c:58

but find this in the X startup log.

(EE) MGA: Failed to load module "mga_hal" (module does not exist, 0) (EE) MGA(0): [agp] Out of memory (-1014)
(EE) MGA(0): [drm] failed to remove DRM signal handler DRIUnlock called when not locked

ideas?

Ed Tomlinson

This one is due to the way AGPGART does (or has done for the last 3 years) its module locking. It does a MOD_INC_USE_COUNT as soon as someone calls the acquire routines. (So you can't unload agpgart whilst you've a 3d using app (like X) open).
This seems quite sensible, but these days you can't unload agpgart.ko anyway because the chipset module (via-agp.ko in your case) already has it 'in use', so I'm tempted to drop those bits.

That's matrox's binary only X blob. Not my fault.

This one is. But it may be a knock-on effect from the bug above. I'll nail that one first.

That one's a problem for the DRI folks.

Dave

| Dave Jones.        http://www.codemonkey.org.uk

Know all about this one... My G400 runs just fine without with X setups.

Thanks

Yep

Now for something new. With bk current (6pm EST) I get:

Dec 21 23:30:56 oscar kernel: Linux agpgart interface v0.100 (c) Dave Jones
Dec 21 23:30:56 oscar kernel: agpgart: Detected VIA MVP3 chipset
Dec 21 23:30:56 oscar kernel:  printing eip:
Dec 21 23:30:56 oscar kernel: e0db9000
Dec 21 23:30:56 oscar kernel: Oops: 0000
Dec 21 23:30:56 oscar kernel: CPU:    0
Dec 21 23:30:56 oscar kernel: EIP:    0060:[<e0db9000>]    Not tainted
Dec 21 23:30:56 oscar kernel: EFLAGS: 00010296
Dec 21 23:30:56 oscar kernel: EIP is at 0xe0db9000
Dec 21 23:30:56 oscar kernel: eax: 00000000   ebx: c15d8800   ecx: 000000a4   edx: 07000203
Dec 21 23:30:56 oscar kernel: esi: 00000000   edi: c15d884c   ebp: c9fc3ee0   esp: c9fc3ec8
Dec 21 23:30:56 oscar kernel: ds: 0068   es: 0068   ss: 0068
Dec 21 23:30:56 oscar kernel: Process modprobe (pid: 2175, threadinfo=c9fc2000 task=d60e92a0)
Dec 21 23:30:56 oscar kernel: Stack: e0de9108 c15d8800 00000000 c15d884c 00000000 00000000 c9fc3ef0 e0de92f8
Dec 21 23:30:56 oscar kernel:        c15d8800 c15d8800 c9fc3f04 e0dd035a c15d8800 e0dcb4a8 a0dcb4a8 c9fc3f28
Dec 21 23:30:56 oscar kernel:        c019d252 c15d8800 e0dd046c c15d884c e0dcb4a8 ffffffed c15d8800 e0dcb480
Dec 21 23:30:56 oscar kernel: Call Trace:
Dec 21 23:30:56 oscar kernel:  [<e0de9108>] agp_backend_initialize+0x1c/0x168 [agpgart]
Dec 21 23:30:56 oscar kernel:  [<e0de92f8>] agp_register_driver+0x2c/0xac [agpgart]
Dec 21 23:30:56 oscar kernel:  [<e0dd035a>] agp_via_probe+0x62/0x6c [via_agp]
Dec 21 23:30:56 oscar kernel:  [<e0dcb4a8>] agp_via_pci_driver+0x28/0xa0 [via_agp]
Dec 21 23:30:56 oscar kernel:  [pci_device_probe+70/96] pci_device_probe+0x46/0x60
Dec 21 23:30:56 oscar kernel:  [<e0dd046c>] agp_via_pci_table+0x0/0x38 [via_agp]
Dec 21 23:30:56 oscar kernel:  [<e0dcb4a8>] agp_via_pci_driver+0x28/0xa0 [via_agp]
Dec 21 23:30:56 oscar kernel:  [<e0dcb480>] agp_via_pci_driver+0x0/0xa0 [via_agp]
Dec 21 23:30:56 oscar kernel:  [bus_match+56/108] bus_match+0x38/0x6c
Dec 21 23:30:56 oscar kernel:  [driver_attach+66/108] driver_attach+0x42/0x6c
Dec 21 23:30:56 oscar kernel:  [<e0dcb4a8>] agp_via_pci_driver+0x28/0xa0 [via_agp]
Dec 21 23:30:56 oscar kernel:  [<e0dcb4a8>] agp_via_pci_driver+0x28/0xa0 [via_agp]
Dec 21 23:30:56 oscar kernel:  [bus_add_driver+172/204] bus_add_driver+0xac/0xcc
Dec 21 23:30:56 oscar kernel:  [<e0dcb4a8>] agp_via_pci_driver+0x28/0xa0 [via_agp]
Dec 21 23:30:56 oscar kernel:  [<e0dcb4c8>] agp_via_pci_driver+0x48/0xa0 [via_agp]
Dec 21 23:30:56 oscar kernel:  [driver_register+54/56] driver_register+0x36/0x38
Dec 21 23:30:56 oscar kernel:  [<e0dcb4a8>] agp_via_pci_driver+0x28/0xa0 [via_agp]
Dec 21 23:30:56 oscar kernel:  [pci_register_driver+68/80] pci_register_driver+0x44/0x50
Dec 21 23:30:56 oscar kernel:  [<e0dcb4a8>] agp_via_pci_driver+0x28/0xa0 [via_agp]
Dec 21 23:30:56 oscar kernel:  [<e0dd0372>] agp_via_init+0xe/0x44 [via_agp]
Dec 21 23:30:56 oscar kernel:  [<e0dcb480>] agp_via_pci_driver+0x0/0xa0 [via_agp]
Dec 21 23:30:56 oscar kernel:  [sys_init_module+274/408] sys_init_module+0x112/0x198
Dec 21 23:30:56 oscar kernel:  [syscall_call+7/11] syscall_call+0x7/0xb
Dec 21 23:30:56 oscar kernel:
Dec 21 23:30:56 oscar kernel: Code:  Bad EIP value.

when modprobing via_agp

Ed

I already fixed a bug with the same call-trace. This looks like you've still got old .o files around. Can you make clean and rebuild just to make sure ?

Dave

| Dave Jones.        http://www.codemonkey.org.uk

Rebuilt with make clean - I am getting the same oops...

Ed

In message <[email protected]> you write:

s

r 0

afe usage in drivers/char/ag

Which is racy under SMP, and under preempt, which is why it's deprecated.

If this is true (it usually is), you can simply drop them. There are other cases where the caller is not grabbing references, so MOD_INC_USE_COUNT is better than nothing (should the warning stay for 2.6? Good question).

Hope that helps,
Rusty.
--
Anyone who quotes me in their sig is an idiot. -- Rusty Russell.

Crapola. I've just realised why this is no longer relevant. I've moved what this was protecting into the per chipset modules. Right now its possible to load modules, start x (which loads DRM), then rmmod via_agp from under its feet. result - boom when something tries to use 3d.

Sure it's unlikely someone would be crazy enough to try and do this, and they deserve what they get, but it's not exactly clean, or nice.

So where is the documentation describing module locking de jour ?

I'll need 'something' in the chipset drivers. The first thing that jumps to mind is to give the chipset drivers an 'acquire' op which does the locking much like the old agp_backend_acquire() does.

Why exactly isn't it safe any more? If there's documentation on this, I'd love to read it. If there isn't, there really should be.

Dave

| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs

In message <[email protected]> you write:

Here's the FAQ again. Note that the init stuff is not currently true (you can enter a module doing init, with possibly bad results), because Linus patched my code. It's on my TODO list to revisit this issue.

The problem is, a module basically can't lock itself. Hence you should expose an "owner" field when you register an interface, and have the caller do the locking. There are other possibly solutions, but none of them were as simple as "lock down before you call in".

Consider the fictitious my_module.c:

	static void my_function_called_through_ptr(void)
	{
		MOD_INC_USE_COUNT;
		...

		MOD_DEC_USE_COUNT;
		return;
	}

If you get preempted before the MOD_INC_USE_COUNT, an unload could occur, boom. If you get preempted after the MOD_DEC_USE_COUNT, but before the return, and an unload occurs, boom.

If it were not called through a pointer, (ie. called by name) then it means the module calling it has a reference (this is done automatically in symbol resolution, or symbol_get()), so no problem.

[ As an implementation detail, this is *not* a problem on SMP without
preemption. But that's an implementation detail, and it wasn't true with the previous module implementation. ]

Hope this helps,
Rusty.
--
Anyone who quotes me in their sig is an idiot. -- Rusty Russell.

Golden Rule: If you are calling though a function pointer into a (different) module, you must hold a reference to that module. Otherwise you risk sleeping in the module while it is unloaded.

Q: How do I get a reference to a module?
A: Usually, a successful call to try_module_get(owner). You don't
need to check for owner != NULL, BTW.

Q: When does try_module_get(owner) fail?
A: When the module is not ready to be entered (ie. still in
init_module) or it is being removed. This prevents you entering the module as it is being discarded (init might fail, or it's being removed).

Q: But the modules' init routine calls my register() routine which
wants to call back into one of the function pointers immediately, and so try_module_get() fails! (because the module is not finished initializing yet)
A: You're being called from the module, so someone already has a
reference (unless there's a bug), so you don't need a try_module_get().

This does mean that if you were to register a structure for *another* module (does anyone do this?) you'd need to have a reference to it.

Q: How do I put the reference back?
A: Using module_put(owner) (owner == NULL is OK).

Q: Do I really need to put try_module_get() before every function ptr
call?
A: If the function does not sleep (any cannot be preempted) ie. is
called in softirq or hardirq context, you can omit this step, since you obviously won't sleep inside the module.

Also, most structs have clear "start" and "stop" functions (eg. mount/umount), so you only need one try_module_get() on start, and module_put() on stop.

Q: Is it safe to call try_module_get() and module_put() from an
interrtupt / softirq?
A: Yes.

Q: My code use "MOD_INC_USE_COUNT". Do I still need to adjust my
module count when someone calls one of my functions?
A: No, you never need to adjust your own module count. There are five
ways a function in your module can get called: firstly, it could be your module_init() function, in which case the module code holds a reference. It could be another module using one of your EXPORT_SYMBOL'ed functions, in which case you cannot be removed since they would have to be removed first. It could be a module which found an EXPORT_SYMBOL'ed function using symbol_get(), in which case they hold a reference count. It could be through a function pointer which your module gave out previously, which is discussed above. Finally, it could be from within your own module, in which case someone must already hold a reference.

Q: My code uses "__MOD_INC_USE_COUNT(reg->owner)", but now I get a
warning at runtime that it is unsafe. What do I need to do?
A: You need to use try_module_get(), and not call into the module if
it fails (act as if it hasn't registered yet). Note that you no longer need to check for NULL yourself, try_module_get() does that.

Q: My code used "GET_USE_COUNT(module)" to get the reference count.
A: Don't do that. If module unloading is disabled, there is no
reference count, and there is never a single value you can assign to.

Q: My code used "try_inc_mod_count(module)" to get the reference
count. Should I change it?
A: No hurry. try_module_get() is exactly the same: the new name
reflects that this is now the only way to get a reference.

Q: How does the code in try_module_get() work?
A: It disables preemption for a moment, checks the live flag, and then
increments a per-cpu counter if the module is live. This is even lighter-weight (in icache and cycles) than using a brlock, but has the same effect. If CONFIG_MODULE_UNLOAD=n, it just becomes a check that the module is live.

Q: How does the module remove code work?
A: It stops the machine by scheduling threads for every other CPU,
then they all disable interrupts. At this stage we know that noone is in try_module_get(), so we can reliably read the counter. If zero, or the rmmod user specified --wait, we set the live flag to false. After this, the reference count should not increase, and each module_put() will wake us up, so we can check the counter again.

Q: Are these changes all so you could implement an in-kernel module
linker?
A: No, they were to prevent load and unload races without altering
every module, nor introducing drastic new requirements.

Q: Doesn't putting linking code into the kernel just add bloat?
A: The total linking code is about 200 generic lines, 100
x86-specific lines. The ia64 linking code is about 500 lines (it's the most complex). Richard Henderson has a great suggestion for simplifying it furthur, which I'm implementing now. "insmod" is now a single portable system call, meaning insmod can be written in about 20 lines of code.

The previous code had to implement the two module loading system calls, the module querying system call, and the /proc/ksyms output, required a little more code than the current x86 linker.

Q: Why not just fix the old code?
A: Because having something so intimate with the kernel in userspace
greatly restricts what changes the kernel can make. Moving into the kernel means I have implemented modversions, typesafe extensible module parameters and kallsyms without altering userspace in any way. Future extensions won't have to worry about the version of modutils problem.

Q: Why not implement two-stage insert / two-stage delete?
A: Because I implemented it first and it sucked. And because this
*is* two-stage insert and two-stage delete, without exposing it to the modules using it, with the added advantage that the second stage is atomic (activation/deactivation is simply changing mod->live, modulo locking implementation magic detailed above). This prevents the race between deactivating the module and finding that someone has starting using it as you are deactivating it.

Hi,

Rusty, you should do this rather soon. Could you please explain, how you intend to solve the module init races? The "don't enter module until init is finished" concept doesn't work, so what else do you want to do?

Q: But if I have a large number of modules (e.g. netfilter) and I have to use try_module_get() for each of them, won't it still affect performance?

Q: Wasn't it possible to do the rewrite in several steps? What made it so urgent that everything had to be 2.6?

2.4.20: wc kernel/module.c include/linux/module.h

   1283    3800   29616 kernel/module.c
    415    1609   13173 include/linux/module.h
   1698    5409   42789 total

2.5.52: wc kernel/module.c kernel/params.c arch/i386/kernel/module.c include/linux/module*
1414 4400 35284 kernel/module.c

    338    1093    8182 kernel/params.c
    124     412    3347 arch/i386/kernel/module.c
    368    1174    9954 include/linux/module.h
     61     244    1958 include/linux/moduleloader.h
    127     619    5199 include/linux/moduleparam.h

2432 7942 63924 total

Q: Doesn't the querying system call provide the same information as /proc/modules and /proc/ksyms? Is procfs now required to load modules?
Q: How can I debug modules with ksymoops?
Q: Why was it necessary to remove the system calls? Wasn't it possible to emulate them?

Q: Can I see the modversions implementation?
Q: Was there an analysis, which discussed the limitation of the old code? Was it really impossible to design a more flexible interface, which left as much as possible in userspace?

Rusty, could you _please_ start answering questions instead of ignoring them or is there no discussion required anymore, now that your code is in the kernel?

bye, Roman