he chain will reject
> big control urbs.
But you did want it for something, didn't you? Or just idle curiosity?
If you're writing microcode, please avoid using control transfers for
uploading firmware without breaking it up explicitly.
-- Pete
-
To unsubscribe from this list: send the li
Two things:
- mbus can be NULL (in case of bus removal while reader is reading)
- Remove a useless assignment
Signed-off-by: Pete Zaitcev <[EMAIL PROTECTED]>
diff -urp -X dontdiff linux-2.6.23-rc3-gregkh/drivers/usb/mon/mon_main.c
linux-2.6.23-rc3-gregkh-pipe/drivers/usb/mon/mon_
Setup packet must be visible in virtual space. There's absolutely no
good reason to implement any kind of zero-copy transfer of 8 bytes, and
the documentation in usb.h is explicit about it. So, drop DMA remapping.
Signed-off-by: Pete Zaitcev <[EMAIL PROTECTED]>
diff -urp -X dontdiff l
This is a set of small updates to Alan's work to make the code more to
my liking. Mostly premature optimizations, but also direction of control
transfers in the binary interface was always out.
Signed-off-by: Pete Zaitcev <[EMAIL PROTECTED]>
diff -urp -X dontdiff linux-2.6.23-rc3-gre
This is a small bunch of cosmetic fixes:
- Timeout is not a write timeout anymore, rename
- Condition in poll was confusingly backwards, invert and simplify
- The comment log gave a wrong impression of version 0.13, terminate it.
Signed-off-by: Pete Zaitcev <[EMAIL PROTECTED]>
---
Add a mutex to protect the ->statusbuf. Not really an issue, because CUPS
is single-threaded when it talks to the printer, but I feel safer this way.
This should be deadlock-free, but I kept this as a separate patch in case
someone ends running a git bisect.
Signed-off-by: Pete Zaitcev <
ers do not?
-- Pete
-
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems? Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of
said that none of his
printers work, which sounds bad, and was news to me. But the point is,
we are trying to extrapolate from too few cases.
I am wondering if Ubuntu has better user reporting, so if Matthew
complains, it really means something.
-- Pete
---
at the suspend methods
for the printer and it seemed possible that they can only work
if the device is not open when suspend occurs. I was going to look
at that at the first opportunity.
So, I'd rather keep at least some kind of capability to suspend
printers, even if disa
ork. I'm wondering if this has something to do with
a hub or motherboard... How should we test it? Tried plugging elsewhere?
-- Pete
-
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find pro
ned out to be a normal,
non-EPiC based Edgeport, and so I cannot test after all. I still
can make a patch, but cannot be sure I got it right.
-- Pete
-
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log fi
return qtd;
> + }
Why use list_for_each_safe when mere list_for_each would do?
In fact, some might ask for list_for_each_entry.
-- Pete
-
This SF.net email is sponsored by: Splunk Inc.
Still grepping
ight? There's no
good reason for setup_packet not to be mapped, right?
-- Pete
-
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems? Stop.
Now Search log events and configu
break;
> }
I am afraid that drivers call unlink left and right, even on URBs which
were not submitted and thus have ->ep == NULL. But on the other hand,
maybe we want to catch them...
-- Pete
-
This SF
On Mon, 30 Jul 2007 13:07:02 -0700, Greg KH <[EMAIL PROTECTED]> wrote:
> On Mon, Jul 30, 2007 at 12:58:57PM -0700, Pete Zaitcev wrote:
> > Oliver converted CmdUrbs into an atomic with commit
> > 96c706ed1c46470598d785124b2a7fb233b27dab .
> > It seems like a waste
Oliver converted CmdUrbs into an atomic with commit
96c706ed1c46470598d785124b2a7fb233b27dab .
It seems like a waste to do for a long forgotten debugging aid.
Let's just remove it altogether.
Signed-off-by: Pete Zaitcev <[EMAIL PROTECTED]>
diff --git a/drivers/usb/serial/io_edgeport.
cle.c
It's a last resort technique which served me well in the past. But
usually if it comes to that it's time to seek a new place of employment,
because obviously you do not have resources in place to see the project
through success.
-- Pete
---
ed to USB_SPEED_FULL
> doesn't work as the packet does not even get sent by the
> subsystem/hcd.
How do you know that the packet is not sent?
The EPROTO you are getting is what happens when packet IS sent and
gets no reply.
-- Pete
(serial, port, cflag & CBAUD);
> + mct_u232_set_baud_rate(serial, port, tty_get_baud_rate(port->tty));
Thanks for doing this, Alan. I noticed that you were on it and wanted
to help but I misunderstood the nature of the change. I thought you
wanted to get rid of switch()... Now I s
On Wed, 25 Jul 2007 08:30:40 +0200, Wolfgang Mües <[EMAIL PROTECTED]> wrote:
> Pete,
>
> On Dienstag, 24. Juli 2007, Pete Zaitcev wrote:
> > Please keep in mind that all this discussion about short reads has
> > nothing to do with your real problem (e.g. the device whi
qh->hw_token and uses it
for length calculation against qtd->length.
-- Pete
-
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems? Stop.
Now Search log events and confi
s a little awkward.
Your new solution is fine, I think. I was concerned that I missed
something so obvious. I rescanned the patch-2.6.23-rc1 looking for
->status being used _outside_ of a callback and didn't see anything
el
on where to
> go searching through the driver code for a solution?
I'm going to double-check that usbmon is not deceiving you. It sure
looks totally bogus. Maybe I mixed up fields in a union somewhere.
-- Pete
-
This S
int retval;
> - int status = urb->status;
>
> - init_completion(&done);
> - urb->context = &done;
> + init_completion(&ctx.done);
> + urb->context = &ctx;
> urb->actual_length
for maximum packet size, and then receive the
actual number of bytes, or even a zero.
Please keep in mind that all this discussion about short reads has
nothing to do with your real problem (e.g. the device which used to
reply with a stall stopped to do so, or it continues to reply with a
;t ask Pavel to run usbmon because I was concerned for EHCI.
But yes, it's normally the first thing to be done because it gives
us some baseline.
-- Pete
-
This SF.net email is sponsored by: Splunk Inc.
Still grepping thr
stupid.
>
> The patch for the kernel:
> http://80211libre.org/at76/at76_usb.patch
I don't see anything suspicious in the patch.
-- Pete
-
This SF.net email is sponsored by: Splunk Inc.
Still grepping through l
Employ the new API URB_FREE_BUFFER that we've got. There was talk of a combined
constructor for this case, but apparently it's not happening, so just set the
flag explicitly for now.
Signed-off-by: Pete Zaitcev <[EMAIL PROTECTED]>
--- linux-2.6.23-rc1-nip/drivers/usb/class/usbl
is.
It has several components; those which invoke LPABORT work, but some of
them need the ioctl added. This is completely compatible, you can mix
old CUPS and new kernels or vice versa.
Signed-off-by: Pete Zaitcev <[EMAIL PROTECTED]>
--- a/drivers/usb/class/usblp.c
+++ b/drivers/usb/class/usblp.
s in the tree.
That said, we can help Alan and Co. if something gets overlooked.
If gigaset or ub stop break, you and I will fix them up.
The goal is to get rid of drivers referencing urb->status. It's going
to be passed as an argument.
-- Pete
-
y definition.
So, I see no opportunity to invoke unlink_urb _here_ as you said.
-- Pete
-
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems? Stop.
Now Search log events and confi
ctures since it's not 32 bits aligned, so I think this attribute
> is correct.
Oliver is correct, everything is perfectly aligned and the attribute
is unnecessary.
However the question is moot, since the bit fields vary in order.
Bit fields must not be used, and shifts and masks must be u
> > [...]
It would help a lot if you trimmed your replies properly.
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limit
appens as long as SLAB
debugging is not enabled. If you want alignment to 256, either
grab pages or realign by hand like ymfpci did.
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version
#x27;s not a delay.
> Any ideas? Btw, yes I've realized the rcvctrlpipe problem. Thanks.
Make sure you aren't doing DMA to stack. The ded0a8c0 does not seem
like an x86 address to me. You might be on an ARM or something.
-- Pete
---
MEOUT);
Are you sure you want to use input pipe with SET_IDLE?
> I've been able to do all of this using libusb in user space
> but have no success within a kernel module.
Compare usbmon traces for two cases, the difference should be evident.
-- Pete
--
y about that.
-- Pete
--- Begin Forwarded Message ---
From: Pete Zaitcev <[EMAIL PROTECTED]>
To: [EMAIL PROTECTED]
Cc: [email protected], [EMAIL PROTECTED]
Subject: usblp: "Big cleanup" breaks O_NONBLOCK
Date: Tue, 10 Jul 2007 20:09:58 -0700
I found the firs
hat the transfer has succeeded even if URB was
merely queued. It is the same behaviour as with the old version.
Signed-off-by: Pete Zaitcev <[EMAIL PROTECTED]>
--- linux-2.6.22-gregkh/drivers/usb/class/usblp.c 2007-07-10
19:40:48.0 -0700
+++ linux-2.6.22-gregkh-eagain/drivers/usb/class
for you even if you tried to open a slice.
However, you cannot do it on dynamic /dev because udev hasn't created
the slice nodes, and it won't do it until partitions are scanned.
Embedded devices usually just run static /dev for that purpose.
Heck, some even use ub with static /dev. The
legal for
HID devices to bunch reports together.
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it no
t you write is true in a way, in case of swiotlb.
The swiotlb reserves a real chunk of memory (still cached though).
That memory remains unavailable to normal users regardless
of the number and size of I/Os currently active in the system.
-- Pete
--
2.6.16.18, it works for
> > both USB2.0 and 1.1.
This looks like an attempt to submit a bulk URB to an interrupt endpoint,
with garbage interval or something of that nature.
I think it would be the best to talk to the driver author directly.
-- Pete
int
address instead of the pipe, because the pipe type is known to each
wrapper, right?
BTW, you sure like empty lines...
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 e
right away. Marcel, please let me know if I'm doing it wrong.
-- Pete
--- a/drivers/usb/class/usblp.c
+++ b/drivers/usb/class/usblp.c
@@ -330,9 +330,6 @@ static void usblp_bulk_write(struct urb *urb)
wake_up(&usblp->wwait);
spin_unlock(&usblp->lock);
nd see
how that goes. If it fails, disable the autosuspend. Although, what
to do if it crashes the device? We'd need to forward this knowledge
to the newly reconnected device somehow, maybe with some kind of a
dynamic blacklist...
-- Pete
not -ERESTARTSYS ?
> >
> > Hmm. Yes, the syscall is restartable at this point. I guess I inherited
> > it from the old driver, where the loop was unified, so it didn't try
> > to distinguish the cases when something was written or nothing was written,
> > and retu
en just as it
always did (e.g. a quick loop-around and a EIO).
> > + if (mutex_lock_interruptible(&usblp->wmut)) {
> > + rv = -EINTR;
>
> Why not -ERESTARTSYS ?
Hmm. Yes, the syscall is restartable at this point. I guess I inherited
it from the old driver, where
missing wake_up(), but I don't know.
Please let me know if you uncover something new and/or come with a patch.
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2
Vojtech agreed to pass usblp over to me, so if you find bugs don't bug him.
Signed-off-by: Pete Zaitcev <[EMAIL PROTECTED]>
Signed-off-by: Vojtech Pavlik <[EMAIL PROTECTED]>
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -3702,12 +3702,12 @@ L: [email protected]
ht
to have them interleaved). Also gone are issues with interrupts using
barriers dangerously.
This patch makes use of Oliver's anchor, and so it must trail the anchor
patch on the way to Linus.
Signed-off-by: Pete Zaitcev <[EMAIL PROTECTED]>
--- linux-2.6.22-rc5-gregkh-usblp/drivers/
Blacklist the hub by VID:PID (0b97:7762)?
But where, in hub.c?
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just
actual problem ("my printer does not work")
instead of reporting the message ("on noes, it's -71!"). Now you see
how usbmon is central to the overarching plan for world domination.
-- Pete
-
This SF.net
more bytes in struct urb ;-)
I think your irony falls flat. Why do you think I haven't submitted
a patch to add it yet? Of course there's always a tradeoff and linear
URBs served us well so far. I am merely stating the reaso
fashion... V.strange.
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourcefo
a hook, we can pass status to it as an argument,
if that helps. I don't expect any trouble from the usbmon's direction.
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE v
ce Association Descriptors in this
> + * configuration. */
> + struct usb_interface_assoc_descriptor *intf_assoc[USB_MAXIADS];
What is this for? How do you want to use this?
-- Pete
-
This SF.net email is sponsored by DB2 E
e. Your hunch that
we'll win anything in driver code is not any better substantiated than
my hunch that URBs are too big and cause memory pressure in unfortunate
moments (otherwise, why does usb-storage continue to lock up with
write throttling issues, but ub works in same tests).
So, another reas
say "Ewww one more field to already-bloated URB"
but if we follow through on making status an argument, no problem.
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2
On Fri, 8 Jun 2007 13:47:54 -0700, Greg KH <[EMAIL PROTECTED]> wrote:
> Thanks, do you think this should go in before 2.6.22 is released?
Considering that we had this bug for 5 years, 2 months won't make
a huge difference. It's not worth any extra eff
fixing as well.
Signed-off-by: Pete Zaitcev <[EMAIL PROTECTED]>
--- a/drivers/usb/core/hcd.c
+++ b/drivers/usb/core/hcd.c
@@ -901,17 +901,32 @@ EXPORT_SYMBOL (usb_calc_bus_time);
/*-*/
-static void urb_unlink (struct ur
idn't know how to correct it.
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
ion under which
it can fail? The patch was completely compliant with the format,
as much as I can see. All the discussion was below the tripple dash.
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 Express C
On Fri, 8 Jun 2007 11:32:32 -0700, Greg KH <[EMAIL PROTECTED]> wrote:
> > and from drivers/usb/core/usb.c:
> >
> > /* format to disable USB on kernel command line is: nousb */
> > __module_param_call("", nousb, param_set_bool, param_get_bool, &n
it with old API
like they do now. The skbuffs are set up this way and it seems
to work.
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of y
nd a compare, and at
worst we'll get a false negative. However, modern compilers optimize
so much that I don't trust anything anymore.
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 E
ction that take
> care of DMA is a good thing.
Just as I suspected.
You know, if you used kmalloc instead, usb_setup_bulk_urb would be
useful for my update of usblp.c.
-- Pete
-
This SF.net email is sponsored by DB2 Ex
sb_buffer_alloc due to confusion. I'm actually wondering now
if your driver is one of the latter...
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and ta
is assertion makes no sense to me. URBs are not "freed" or "put" by
magic by the core when your completion handler returns, thank heavens!
Your own completion handler calls usb_free_urb, when the status is known.
So, if you need to
when we read them in?
I remember that issue. I thought that we wanted some kind of escape
syntax... Like what HTML uses with &#; perhaps. This would allow
to edit xorg.conf on systems which are not UTF-8 clean. But perhaps
it's a non-goal. How big is the code to convert (we
ROTECTED] ~]$ cat /etc/udev/rules.d/70-gps.rules
KERNEL=="ttyACM*", SYSFS{idVendor}=="0b20" SYSFS{idProduct}=="0406",
SYMLINK+="gps0"
[EMAIL PROTECTED] ~]$
-- Pete
-
This SF.net email is
Release DMA resources if submission fails in the HCD.
Signed-Off-By: Pete Zaitcev <[EMAIL PROTECTED]>
---
> Amazing indeed.
Just for the record, I didn't find it just by looking. IBM filed a bug
RH#236922 after they ran out of IOMMU. Our (older) uhci-hcd was rejecting
bulk re
Release DMA resources if submission fails in the HCD.
Signed-Off-By: Pete Zaitcev <[EMAIL PROTECTED]>
---
This seems too obvious. Yet I went back to kernel 2.2.19 and we always
had this bug (well, in 2.2 we didn't because we used virt_to_bus).
I'm amazed that such things can l
!= (head);
>pos = n, n = list_entry(n->member.next, typeof(*n), member))
>
> then what am I to think?
You won't catch me writing this kind of crap, so the question is moot.
Seriously, a comma operator? Admit it, you just expanded a marcro from
list.h by
not find a better way to blow up your own argument
than to mention list_for_each_entry_safe(), which is anything but.
Matthias' use of list_for_each_entry() actually IS safe, which is
why I am not NAKing it. Andrew has
working knowledge of them. Iterators are
particularly nasty that way. I'm thinking about dropping all of these
list_for_each_with_murky_argument_requirements_and_odd_side_effects()
and use plain for(;;), as a courtesy to someone who has to read the
code years down the road.
-- Pete
ch straddles the border of acceptable. The pointless obfuscation
is balanced against the removal of explicit type in list_entry() and thus
a possible mismatched struct. I'm not acking nor naking this.
-- Pete
-
This SF.net e
Suspend destroys refcounting for open/release.
Signed-off-by: Pete Zaitcev <[EMAIL PROTECTED]>
---
This whole driver needs a good beating, I swear.
diff --git a/drivers/usb/class/usblp.c b/drivers/usb/class/usblp.c
index fc2624d..2c74880 100644
--- a/drivers/usb/class/usblp.c
+++ b/drive
ed and lame, other people love it. If someone (Laurent?)
rolls out a buffer allocator which builds upon vmalloc, I guess
I'll have to look for specific flaws in that. It's not something
completely impossible, but we'l
Looks like the error path had a copy-paste error. The normal exit path
uses correct URB already.
Signed-off-by: Pete Zaitcev <[EMAIL PROTECTED]>
--- a/drivers/usb/class/usblp.c
+++ b/drivers/usb/class/usblp.c
@@ -1004,7 +1004,7 @@ abort:
usblp->writeb
. There,
we solved the issue by having a compilation unit (pci-quirks.c)
which textually belongs to the driver, yet is built into kernel.
What I proposed is more like how the handoff quirk worked before.
But if you know an sane alternative, I'll be happy (ok, Mar
ore part. Fill the word
when ohci probes, let ehci use it. Then, see above.
The patch was linewrapped, the coding style was horrible, but
it was all secondary to the overall architecture. Once you get
rid of inter-module interaction, we can look at small things.
They are easy.
Best wishes,
-- Pete
-
one, which has
> always been kind of dubious and acts unmaintained). And that routine
> doesn't do a local_irq_save.
Absolutely right, I'm being an idiot here. I think I looked at root
hub code in the rush to the FreedomHEC preparations. We do not have
local_irq_save in th
rm allows it,
this is done by masking a specific IRQ. If that is impossible, interrupts
are masked in the CPU (maybe only partially, e.g. on SPARC). However
it is done though, the local_irq_save that we do adds nothing that
I can see.
-- Pete
;s interrupt handling, and we'll only
need to make it explicit if any HCDs start using softirq when
calling the giveback routine). I'm poking your memory in case
there's actually a good reason for it which I forgot.
Yours,
-- Pete
---
rnel
hacker) receive at least some benefit. If I turned Hans away right
off the bat, his cohorts would still do the same thing, only without
talking to us. That would be a net loss for Linux.
-- Pete
-
This SF.net email is spons
> of "usb_submit_urb()" after that the device has detached! It must be
> something terribly wrong in the Linux USB stack if the callbacks are alive
> after that you have detached a USB device.
I'm so blogging this.
-- Pete
-
submission, it is reported
on the context which is submitting. Thus, you cannot hold any
locks across the submission which the callback is likely to take.
It's a real pain on SMP and with a multithreaded system.
-- Pete
-
T
ties by default and only use them when
> >memory pressure invokes these URBs
> >
> > This is something which, I think, only Greg Kroah can decide on doing
> > or not doing.
>
> I am not in favor of such a large change.
Yeah...
I mentioned in m
, but it cannot wait for
a writeout. If it cannot be satisfied without writing dirty data, it
will fail.
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and t
pported on Linux. Heck it works
better than the equivalent on x86_64, because there's no address range
limitation. The key is to set IOMMU at transfer time. Solaris defers
DMA setup as well (for what they call "streaming transfers").
-- Pete
--
scsi_eh_0, and usb-storage helper.
I am thinking about taking the ub concept and integrating it into
usb-storage as some kind of "threadless" option. If only I could
get rid of the SCSI error handling...
-- Pete
-
r, less
complex, less buggy, used commonly, and another far more complex,
only used by usb-storage. Personally, I don't see it flying.
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the F
quick glance.
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/
e old driver did here, sorry.
But oddly enough, it only returned ENOMEM for writes, but not reads.
I changed this to match the legacy.
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE ver
e in drivers/char/lp.c
- Greg-style dynamic URBs
-- Pete
P.S. I'm thinking if I should take usblp for maintenance from Vojtech
since I am reworking it anyway.
P.P.S. Does anyone remember what hash signs in front of copyright
statements mean?
diff --git a/drivers/usb/class/usblp.c b/driver
got its name) to fetch the
current SOF and then call gettimeofday? Rest assured that the jitter
you get from scheduling is going to be there for any in-kernel
implementation as well.
-- Pete
-
This SF.net email is sponsored b
want
to make sure that a scheduled work won't be run when the module is gone.
More often, a device removal is the issue, but as I take it, such is not
possible for a built-in device :-) . In most cases, all it takes is a
strategically placed flu
more unconditionally regardless of hardware present seems like
stepping over the line to me.
Overall this seems like a good patch to me, only needs to be less
intrusive for owners of other hardware.
-- Pete
-
This SF.net email is
a host to keep wider counters
than are needed.
-- Pete
-
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to
wn timestamp is not good enough,
because the device does not know it. Actually, I think that a clever
design would set an interrupt-out transfer with software initiated
timestamp, which would not depend on HC implementing SOF number
as a monotonously increasing counte
1 - 100 of 1241 matches
Mail list logo
