On Fri, 6 Jan 2006 18:38:46 +, Russell King wrote:
>> is there any architecture for which packed is required in structures like
>> this:
>>
>> /* All standard descriptors have these 2 fields at the beginning */
>> struct usb_descriptor_header {
>> __u8 bLength;
>> __u8 bDescriptor
On Thu, Jan 12, 2006 at 01:27:12PM +0100, Mikael Pettersson wrote:
> On Fri, 6 Jan 2006 18:38:46 +, Russell King wrote:
> >> is there any architecture for which packed is required in structures like
> >> this:
> >>
> >> /* All standard descriptors have these 2 fields at the beginning */
> >>
On Thu, Jan 12, 2006 at 01:47:29PM +, Russell King wrote:
> Due to lack of manpower on the Linux side (iow, more or less just me)
> this became the ABI of the early ARM Linux a.out toolchain. At that
> time, I did not consider this to be a problem - it wasn't a problem
> as far as the kernel w
Russell King writes:
> > As fas as I can tell, the AAPCS document (v2.03 7th Oct 2005) requires
> > that a simple "struct foo { unsigned char c; };" should have both size
> > and alignment equal to 1, but gcc makes them both 4. Do you have any
> > information about why gcc is doing this on ARM/
On Thu, Jan 12, 2006 at 05:30:11PM +0100, Mikael Pettersson wrote:
> OK, thanks for this info. It means that GCC is the definitive authority
> on calling conventions and data layouts, not the AAPCS; I wasn't aware of
> that before.
BTW, it's worth noting that the new EABI stuff has it's own set of
On Thu, 12 Jan 2006 13:27:12 +0100 (MET), Mikael Pettersson <[EMAIL PROTECTED]>
wrote:
> [...] Do you have any
> information about why gcc is doing this on ARM/Linux?
Russell forgot to explain it, but the reason for this weirdness is real.
It is so you can do things like this:
struct foo {
On Thu, Jan 12, 2006 at 09:20:06AM -0800, Pete Zaitcev wrote:
> P.S. I am repeating myself as Katon, but I am yet to see why any of
> this matters. Neither Russell nor Oliver ever presented a case where
> an unpacked struct caused breakage in USB.
If you would like to refresh your memory (which is
On Thu, 12 Jan 2006 17:26:17 +, Russell King <[EMAIL PROTECTED]> wrote:
> On Thu, Jan 12, 2006 at 09:20:06AM -0800, Pete Zaitcev wrote:
> > P.S. I am repeating myself as Katon, but I am yet to see why any of
> > this matters. Neither Russell nor Oliver ever presented a case where
> > an unpack
On Fri, Jan 06, 2006 at 06:38:46PM +, Russell King wrote:
> > /* All standard descriptors have these 2 fields at the beginning */
> > struct usb_descriptor_header {
> > __u8 bLength;
> > __u8 bDescriptorType;
> > };
>
> sizeof(struct usb_descriptor_header) will be 4 on ARM. If this
On Fri, Jan 06, 2006 at 07:15:43PM +0100, Oliver Neukum wrote:
> Hi,
>
> is there any architecture for which packed is required in structures like
> this:
>
> /* All standard descriptors have these 2 fields at the beginning */
> struct usb_descriptor_header {
> __u8 bLength;
> __u8
Am Freitag, 6. Januar 2006 19:38 schrieb Russell King:
> On Fri, Jan 06, 2006 at 07:15:43PM +0100, Oliver Neukum wrote:
> > Hi,
> >
> > is there any architecture for which packed is required in structures like
> > this:
> >
> > /* All standard descriptors have these 2 fields at the beginning */
On Fri, 6 Jan 2006 14:33:21 -0500 (EST), Alan Stern <[EMAIL PROTECTED]> wrote:
> > > > is there any architecture for which packed is required in structures
> > > > like this:
> > > >
> > > > /* All standard descriptors have these 2 fields at the beginning */
> > > > struct usb_descriptor_header
On Fri, Jan 06, 2006 at 12:23:00PM -0800, Pete Zaitcev wrote:
> Also, even if RMK is right, it either happens with some garbage gcc version
> or it is just a bug which didn't reach gcc people yet.
You're wrong on both counts. It's part of the ABI.
> Why do we all have to
> suffer because of them
On Fri, 6 Jan 2006 22:13:56 +0100, Oliver Neukum <[EMAIL PROTECTED]> wrote:
> These data structures are laid out in natural alignment. Is there anything
> in the gcc documentation? I find a lot about what packed does, but where
> is described what is done without packed?
Russell said it's in ABI
Am Freitag, 6. Januar 2006 22:22 schrieb Pete Zaitcev:
> On Fri, 6 Jan 2006 22:13:56 +0100, Oliver Neukum <[EMAIL PROTECTED]> wrote:
>
> > These data structures are laid out in natural alignment. Is there anything
> > in the gcc documentation? I find a lot about what packed does, but where
> > is
Russell King wrote:
> BTW, it's worth noting that the new EABI stuff has it's own set of
> problems. We have r0 to r6 to pass 32-bit or 64-bit arguments.
> With EABI, 64-bit arguments will be aligned to an _even_ numbered
> register.
Is there a reason for this alignment requirement?
David Vrabel
On Thu, Jan 12, 2006 at 05:22:47PM +, David Vrabel wrote:
> Russell King wrote:
> > BTW, it's worth noting that the new EABI stuff has it's own set of
> > problems. We have r0 to r6 to pass 32-bit or 64-bit arguments.
> > With EABI, 64-bit arguments will be aligned to an _even_ numbered
> > re
On Thursday 12 January 2006, Pete Zaitcev wrote (in another snide postscript):
>
> P.P.S. The USB stack was careful to use correct sizes historically.
> One grep of the source will tell you that all this stench emanates from
> the newer code, in particular the gadget and its attendant components,
On Fri, 6 Jan 2006, Oliver Neukum wrote:
> Am Freitag, 6. Januar 2006 19:38 schrieb Russell King:
> > On Fri, Jan 06, 2006 at 07:15:43PM +0100, Oliver Neukum wrote:
> > > Hi,
> > >
> > > is there any architecture for which packed is required in structures like
> > > this:
> > >
> > > /* All sta
Am Freitag, 6. Januar 2006 20:33 schrieb Alan Stern:
> On Fri, 6 Jan 2006, Oliver Neukum wrote:
>
> > Am Freitag, 6. Januar 2006 19:38 schrieb Russell King:
> > > On Fri, Jan 06, 2006 at 07:15:43PM +0100, Oliver Neukum wrote:
> > > > Hi,
> > > >
> > > > is there any architecture for which packed
Am Freitag, 6. Januar 2006 21:23 schrieb Pete Zaitcev:
> On Fri, 6 Jan 2006 14:33:21 -0500 (EST), Alan Stern <[EMAIL PROTECTED]> wrote:
>
> > > > > is there any architecture for which packed is required in structures
> > > > > like this:
> > > > >
> > > > > /* All standard descriptors have these
On Fri, Jan 06, 2006 at 09:46:09PM +0100, Oliver Neukum wrote:
> in any case the file is inconsistent. Either all structures there have to
> be packed or none. Unless someone can come up with something better
> I am afraid an "#ifdef ARM" is needed. Comments?
I would suggest ignoring Pete - all re
Am Freitag, 6. Januar 2006 21:49 schrieb Russell King:
> On Fri, Jan 06, 2006 at 09:46:09PM +0100, Oliver Neukum wrote:
> > in any case the file is inconsistent. Either all structures there have to
> > be packed or none. Unless someone can come up with something better
> > I am afraid an "#ifdef AR
On Fri, 6 Jan 2006, Oliver Neukum wrote:
> > Perhaps you are overreacting. Even though the total size of the structure
> > may be 4, I bet that bLength and bDescriptorType still occupy just the
> > first two bytes of the structure. Assuming this is so, there's nothing
> > wrong with the exist
Am Freitag, 6. Januar 2006 22:59 schrieb Alan Stern:
> On Fri, 6 Jan 2006, Oliver Neukum wrote:
>
> > > Perhaps you are overreacting. Even though the total size of the
> > > structure
> > > may be 4, I bet that bLength and bDescriptorType still occupy just the
> > > first two bytes of the stru
On Fri, 6 Jan 2006, Oliver Neukum wrote:
> Am Freitag, 6. Januar 2006 22:59 schrieb Alan Stern:
> > On Fri, 6 Jan 2006, Oliver Neukum wrote:
> >
> > > > Perhaps you are overreacting. Even though the total size of the
> > > > structure
> > > > may be 4, I bet that bLength and bDescriptorType st
On Fri, 6 Jan 2006 23:07:59 +0100, Oliver Neukum <[EMAIL PROTECTED]> wrote:
> Sorry. I confused it with usb_device_descriptor which is appended to in
> devio.c
> WHich is packed. Considering pure alignment it shouldn't be. Yet there seems
> to be a problem in arm. Can you shed light on this?
You
Am Freitag, 6. Januar 2006 23:18 schrieb Alan Stern:
> I don't know why struct usb_device_descriptor is packed. In fact, most of
> the structures defined in usb_ch9.h are packed for no apparent reason.
> Maybe it's so you can dereference descriptors embedded in the
> rawdescriptors array; those
On Gwe, 2006-01-06 at 12:23 -0800, Pete Zaitcev wrote:
> or it is just a bug which didn't reach gcc people yet. It is dysfunctional
> to have structures padded in this way. On all other platforms they are
> only padded if they include a member which is bigger than the char.
Not true. Maybe for the
On Fri, 6 Jan 2006, Oliver Neukum wrote:
> Am Freitag, 6. Januar 2006 23:18 schrieb Alan Stern:
> > I don't know why struct usb_device_descriptor is packed. In fact, most of
> > the structures defined in usb_ch9.h are packed for no apparent reason.
> > Maybe it's so you can dereference descript
On Friday 06 January 2006 2:18 pm, Alan Stern wrote:
>
> I don't know why struct usb_device_descriptor is packed. In fact, most of
> the structures defined in usb_ch9.h are packed for no apparent reason.
To ensure not just that the protocol formats are strictly adhered to,
but that it's obviou
On Friday 06 January 2006 12:46 pm, Oliver Neukum wrote:
> in any case the file is inconsistent. Either all structures there have to
> be packed or none. Unless someone can come up with something better
> I am afraid an "#ifdef ARM" is needed. Comments?
Just pack them all. Your patch looked OK t
32 matches
Mail list logo
