Skip to contentSkip to conversation

BeagleBoard using GCC 4.6.0

DISCUSSION Gary Thomas 4 participants 17 messages Jun 8, 2011 – Jun 10, 2011 USB

lore
Share & tools
lore
#0 lore raw

Now that Poky is using GCC 4.6.0, has anyone actually checked the operation on the BeagleBoard? I suspect that you'll find that the EHCI USB does not work. I can't really check here as I don't trust the EHCI on my BeagleBoard rev C3 (not xM).

That said, I've tried this new compiler (previously reported) on my own OMAP/3530 board which uses the same 2.6.37 kernel (just not the linux-yocto version). When I build my kernel with GCC 4.5.2, it works perfectly, but fails with GCC 4.6.0 (I trust the hardware).

I've isolated it down to at least the function 'ehci_hub_control' (but I suspect the problem is more fundamental). Comparing the code generated by the two compilers with the same source tree, this function
is dramatically different. I can see why it's failing, I just don't
know why the compiler is doing what it's doing.

The lines at drivers/usb/host/ehci-hub.c:841

	case GetPortStatus:
		if (!wIndex || wIndex > ports)
			goto error;
		wIndex--;
		status = 0;
		temp = ehci_readl(ehci, status_reg);

are being compiled very differently.

With GCC 4.5.2:

0xc0229810 <ehci_hub_control+672>:      cmp     r3, #0  ; 0x0
0xc0229814 <ehci_hub_control+676>:      beq     0xc0229df8 <ehci_hub_control+2184>
0xc0229818 <ehci_hub_control+680>:      cmp     r3, r0
0xc022981c <ehci_hub_control+684>:      bgt     0xc0229df8 <ehci_hub_control+2184>
0xc0229820 <ehci_hub_control+688>:      sub     r8, r3, #1      ; 0x1
0xc0229824 <ehci_hub_control+692>:      ldr     r5, [r7, #4]
0xc0229828 <ehci_hub_control+696>:      uxth    r8, r8
0xc022982c <ehci_hub_control+700>:      dmb     sy

With GCC 4.6.0:

0xc0221630 <ehci_hub_control+808>:      cmp     r3, #0  ; 0x0
0xc0221634 <ehci_hub_control+812>:      beq     0xc0221d7c <ehci_hub_control+2676>
0xc0221638 <ehci_hub_control+816>:      cmp     r3, r0
0xc022163c <ehci_hub_control+820>:      bgt     0xc0221d7c <ehci_hub_control+2676>
0xc0221640 <ehci_hub_control+824>:      sub     r7, r3, #1      ; 0x1
0xc0221644 <ehci_hub_control+828>:      add     r3, r11, #16    ; 0x10
0xc0221648 <ehci_hub_control+832>:      add     r3, r8, r3, lsl #2
0xc022164c <ehci_hub_control+836>:      uxth    r7, r7
0xc0221650 <ehci_hub_control+840>:      ldrb    r5, [r3, #5]
0xc0221654 <ehci_hub_control+844>:      ldrb    r2, [r3, #4]
0xc0221658 <ehci_hub_control+848>:      orr     r5, r2, r5, lsl #8
0xc022165c <ehci_hub_control+852>:      ldrb    r2, [r3, #6]
0xc0221660 <ehci_hub_control+856>:      ldrb    r3, [r3, #7]
0xc0221664 <ehci_hub_control+860>:      orr     r5, r5, r2, lsl #16
0xc0221668 <ehci_hub_control+864>:      orr     r5, r5, r3, lsl #24
0xc022166c <ehci_hub_control+868>:      dmb     sy

As you can see, the old compiler accesses the ehci status register in a single access, the new compiler dances around and makes multiple accesses, which in the end get very incorrect data (I think that this register is like many which clear bits on reads).

Any ideas where I can go with this? I'm really trying to keep up with Poky/Yocto, but this move to GCC 4.6.0 has broken my ARM targets :-( I do have PowerPC targets as well - they seem fine (from limited testing) with the new compiler.

Note: if you want to see the whole function disassembly, look at
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.5.2
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.6.0

Thanks

Signature · 4 lines
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------
#1 lore raw
Quoted 10 lines · Gary Thomas

Now that Poky is using GCC 4.6.0, has anyone actually checked the operation on the BeagleBoard? I suspect that you'll find that the EHCI USB does not work. I can't really check here as I don't trust the EHCI on my BeagleBoard rev C3 (not xM).

That said, I've tried this new compiler (previously reported) on my own OMAP/3530 board which uses the same 2.6.37 kernel (just not the linux-yocto version). When I build my kernel with GCC 4.5.2, it works perfectly, but fails with GCC 4.6.0 (I trust the hardware).

Quoted 6 lines

I've isolated it down to at least the function 'ehci_hub_control' (but I suspect the problem is more fundamental). Comparing the code generated by the two compilers with the same source tree, this function is dramatically different. I can see why it's failing, I just don't know why the compiler is doing what it's doing.

Gary,

I am not sure if Khem Raj is on this list, so I am forwarding it to him, he might have some insight.

Sau!

Quoted 54 lines

The lines at drivers/usb/host/ehci-hub.c:841
case GetPortStatus:
if (!wIndex || wIndex > ports)
goto error;
wIndex--;
status = 0;
temp = ehci_readl(ehci, status_reg);

are being compiled very differently.

With GCC 4.5.2:
0xc0229810 <ehci_hub_control+672>: cmp r3, #0 ; 0x0 0xc0229814 <ehci_hub_control+676>: beq 0xc0229df8 <ehci_hub_control+2184> 0xc0229818 <ehci_hub_control+680>: cmp r3, r0
0xc022981c <ehci_hub_control+684>: bgt 0xc0229df8 <ehci_hub_control+2184> 0xc0229820 <ehci_hub_control+688>: sub r8, r3, #1 ; 0x1 0xc0229824 <ehci_hub_control+692>: ldr r5, [r7, #4] 0xc0229828 <ehci_hub_control+696>: uxth r8, r8
0xc022982c <ehci_hub_control+700>: dmb sy

With GCC 4.6.0:
0xc0221630 <ehci_hub_control+808>: cmp r3, #0 ; 0x0 0xc0221634 <ehci_hub_control+812>: beq 0xc0221d7c <ehci_hub_control+2676> 0xc0221638 <ehci_hub_control+816>: cmp r3, r0
0xc022163c <ehci_hub_control+820>: bgt 0xc0221d7c <ehci_hub_control+2676> 0xc0221640 <ehci_hub_control+824>: sub r7, r3, #1 ; 0x1 0xc0221644 <ehci_hub_control+828>: add r3, r11, #16 ; 0x10 0xc0221648 <ehci_hub_control+832>: add r3, r8, r3, lsl #2 0xc022164c <ehci_hub_control+836>: uxth r7, r7
0xc0221650 <ehci_hub_control+840>: ldrb r5, [r3, #5] 0xc0221654 <ehci_hub_control+844>: ldrb r2, [r3, #4] 0xc0221658 <ehci_hub_control+848>: orr r5, r2, r5, lsl #8 0xc022165c <ehci_hub_control+852>: ldrb r2, [r3, #6] 0xc0221660 <ehci_hub_control+856>: ldrb r3, [r3, #7] 0xc0221664 <ehci_hub_control+860>: orr r5, r5, r2, lsl #16 0xc0221668 <ehci_hub_control+864>: orr r5, r5, r3, lsl #24 0xc022166c <ehci_hub_control+868>: dmb sy

As you can see, the old compiler accesses the ehci status register in a single access, the new compiler dances around and makes multiple accesses, which in the end get very incorrect data (I think that this register is like many which clear bits on reads).

Any ideas where I can go with this? I'm really trying to keep up with Poky/Yocto, but this move to GCC 4.6.0 has broken my ARM targets :-( I do have PowerPC targets as well - they seem fine (from limited testing) with the new compiler.

Note: if you want to see the whole function disassembly, look at
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.5.2
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.6.0

Thanks

#15 lore raw
In reply to Gary Thomas
Quoted 4 lines · Gary Thomas

Now that Poky is using GCC 4.6.0, has anyone actually checked the operation on the BeagleBoard? I suspect that you'll find that the EHCI USB does not work. I can't really check here as I don't trust the EHCI on my BeagleBoard rev C3 (not xM).

Gary,

Thank you for digging into this. Excellent timing as I was looking at the same thing - I hadn't looked at the compiler yet as I was in the middle of some kernel work and had assumed the problem lay there. I'll see if I can try with the older kernel to get this resolved as well. Have you tried with various optimization options yet?

--
Darren

Quoted 66 lines

That said, I've tried this new compiler (previously reported) on my own OMAP/3530 board which uses the same 2.6.37 kernel (just not the linux-yocto version). When I build my kernel with GCC 4.5.2, it works perfectly, but fails with GCC 4.6.0 (I trust the hardware).

I've isolated it down to at least the function 'ehci_hub_control' (but I suspect the problem is more fundamental). Comparing the code generated by the two compilers with the same source tree, this function
is dramatically different. I can see why it's failing, I just don't
know why the compiler is doing what it's doing.

The lines at drivers/usb/host/ehci-hub.c:841

	case GetPortStatus:
		if (!wIndex || wIndex > ports)
			goto error;
		wIndex--;
		status = 0;
		temp = ehci_readl(ehci, status_reg);

are being compiled very differently.

With GCC 4.5.2:

0xc0229810 <ehci_hub_control+672>:      cmp     r3, #0  ; 0x0
0xc0229814 <ehci_hub_control+676>:      beq     0xc0229df8 <ehci_hub_control+2184>
0xc0229818 <ehci_hub_control+680>:      cmp     r3, r0
0xc022981c <ehci_hub_control+684>:      bgt     0xc0229df8 <ehci_hub_control+2184>
0xc0229820 <ehci_hub_control+688>:      sub     r8, r3, #1      ; 0x1
0xc0229824 <ehci_hub_control+692>:      ldr     r5, [r7, #4]
0xc0229828 <ehci_hub_control+696>:      uxth    r8, r8
0xc022982c <ehci_hub_control+700>:      dmb     sy

With GCC 4.6.0:

0xc0221630 <ehci_hub_control+808>:      cmp     r3, #0  ; 0x0
0xc0221634 <ehci_hub_control+812>:      beq     0xc0221d7c <ehci_hub_control+2676>
0xc0221638 <ehci_hub_control+816>:      cmp     r3, r0
0xc022163c <ehci_hub_control+820>:      bgt     0xc0221d7c <ehci_hub_control+2676>
0xc0221640 <ehci_hub_control+824>:      sub     r7, r3, #1      ; 0x1
0xc0221644 <ehci_hub_control+828>:      add     r3, r11, #16    ; 0x10
0xc0221648 <ehci_hub_control+832>:      add     r3, r8, r3, lsl #2
0xc022164c <ehci_hub_control+836>:      uxth    r7, r7
0xc0221650 <ehci_hub_control+840>:      ldrb    r5, [r3, #5]
0xc0221654 <ehci_hub_control+844>:      ldrb    r2, [r3, #4]
0xc0221658 <ehci_hub_control+848>:      orr     r5, r2, r5, lsl #8
0xc022165c <ehci_hub_control+852>:      ldrb    r2, [r3, #6]
0xc0221660 <ehci_hub_control+856>:      ldrb    r3, [r3, #7]
0xc0221664 <ehci_hub_control+860>:      orr     r5, r5, r2, lsl #16
0xc0221668 <ehci_hub_control+864>:      orr     r5, r5, r3, lsl #24
0xc022166c <ehci_hub_control+868>:      dmb     sy

As you can see, the old compiler accesses the ehci status register in a single access, the new compiler dances around and makes multiple accesses, which in the end get very incorrect data (I think that this register is like many which clear bits on reads).

Any ideas where I can go with this? I'm really trying to keep up with Poky/Yocto, but this move to GCC 4.6.0 has broken my ARM targets :-( I do have PowerPC targets as well - they seem fine (from limited testing) with the new compiler.

Note: if you want to see the whole function disassembly, look at
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.5.2
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.6.0

Thanks

Signature · 3 lines
Darren Hart
Intel Open Source Technology Center
Yocto Project - Linux Kernel
#16 lore raw
Quoted 6 lines · Darren Hart
Quoted 4 lines · Gary Thomas

Now that Poky is using GCC 4.6.0, has anyone actually checked the operation on the BeagleBoard? I suspect that you'll find that the EHCI USB does not work. I can't really check here as I don't trust the EHCI on my BeagleBoard rev C3 (not xM).

Quoted 7 lines

Gary,

Thank you for digging into this. Excellent timing as I was looking at the same thing - I hadn't looked at the compiler yet as I was in the middle of some kernel work and had assumed the problem lay there. I'll see if I can try with the older kernel to get this resolved as well. Have you tried with various optimization options yet?

Not yet, I've only tried the basic options, etc, to try and see where things might be going South.

Quoted 67 lines
Quoted 66 lines

That said, I've tried this new compiler (previously reported) on my own OMAP/3530 board which uses the same 2.6.37 kernel (just not the linux-yocto version). When I build my kernel with GCC 4.5.2, it works perfectly, but fails with GCC 4.6.0 (I trust the hardware).

I've isolated it down to at least the function 'ehci_hub_control' (but I suspect the problem is more fundamental). Comparing the code generated by the two compilers with the same source tree, this function
is dramatically different. I can see why it's failing, I just don't
know why the compiler is doing what it's doing.

The lines at drivers/usb/host/ehci-hub.c:841

	case GetPortStatus:
		if (!wIndex || wIndex>  ports)
			goto error;
		wIndex--;
		status = 0;
		temp = ehci_readl(ehci, status_reg);

are being compiled very differently.

With GCC 4.5.2:

0xc0229810<ehci_hub_control+672>:      cmp     r3, #0  ; 0x0
0xc0229814<ehci_hub_control+676>:      beq     0xc0229df8<ehci_hub_control+2184>
0xc0229818<ehci_hub_control+680>:      cmp     r3, r0
0xc022981c<ehci_hub_control+684>:      bgt     0xc0229df8<ehci_hub_control+2184>
0xc0229820<ehci_hub_control+688>:      sub     r8, r3, #1      ; 0x1
0xc0229824<ehci_hub_control+692>:      ldr     r5, [r7, #4]
0xc0229828<ehci_hub_control+696>:      uxth    r8, r8
0xc022982c<ehci_hub_control+700>:      dmb     sy

With GCC 4.6.0:

0xc0221630<ehci_hub_control+808>:      cmp     r3, #0  ; 0x0
0xc0221634<ehci_hub_control+812>:      beq     0xc0221d7c<ehci_hub_control+2676>
0xc0221638<ehci_hub_control+816>:      cmp     r3, r0
0xc022163c<ehci_hub_control+820>:      bgt     0xc0221d7c<ehci_hub_control+2676>
0xc0221640<ehci_hub_control+824>:      sub     r7, r3, #1      ; 0x1
0xc0221644<ehci_hub_control+828>:      add     r3, r11, #16    ; 0x10
0xc0221648<ehci_hub_control+832>:      add     r3, r8, r3, lsl #2
0xc022164c<ehci_hub_control+836>:      uxth    r7, r7
0xc0221650<ehci_hub_control+840>:      ldrb    r5, [r3, #5]
0xc0221654<ehci_hub_control+844>:      ldrb    r2, [r3, #4]
0xc0221658<ehci_hub_control+848>:      orr     r5, r2, r5, lsl #8
0xc022165c<ehci_hub_control+852>:      ldrb    r2, [r3, #6]
0xc0221660<ehci_hub_control+856>:      ldrb    r3, [r3, #7]
0xc0221664<ehci_hub_control+860>:      orr     r5, r5, r2, lsl #16
0xc0221668<ehci_hub_control+864>:      orr     r5, r5, r3, lsl #24
0xc022166c<ehci_hub_control+868>:      dmb     sy

As you can see, the old compiler accesses the ehci status register in a single access, the new compiler dances around and makes multiple accesses, which in the end get very incorrect data (I think that this register is like many which clear bits on reads).

Any ideas where I can go with this? I'm really trying to keep up with Poky/Yocto, but this move to GCC 4.6.0 has broken my ARM targets :-( I do have PowerPC targets as well - they seem fine (from limited testing) with the new compiler.

Note: if you want to see the whole function disassembly, look at

    http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.5.2
    http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.6.0

Thanks

Signature · 4 lines
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------
#2 lore raw
In reply to Saul Wold
Quoted 18 lines · Saul Wold
Quoted 10 lines · Gary Thomas

Now that Poky is using GCC 4.6.0, has anyone actually checked the operation on the BeagleBoard? I suspect that you'll find that the EHCI USB does not work. I can't really check here as I don't trust the EHCI on my BeagleBoard rev C3 (not xM).

That said, I've tried this new compiler (previously reported) on my own OMAP/3530 board which uses the same 2.6.37 kernel (just not the linux-yocto version). When I build my kernel with GCC 4.5.2, it works perfectly, but fails with GCC 4.6.0 (I trust the hardware).

Quoted 6 lines

I've isolated it down to at least the function 'ehci_hub_control' (but I suspect the problem is more fundamental). Comparing the code generated by the two compilers with the same source tree, this function is dramatically different. I can see why it's failing, I just don't know why the compiler is doing what it's doing.

Quoted 4 lines

Gary,

I am not sure if Khem Raj is on this list, so I am forwarding it to him, he might have some insight.

By switching GCCVERSION to 4.5.1, a kernel that previously failed to bring up the USB hub on the beagleboard xM (and the ethernet port) with 4.6.0, now successfully does so. Unfortunatley, the framebuffer device fails now. Any hints from the compiler crowd on things to try? Or output we can collect to help resolve this?

--
Darren

Quoted 62 lines

Sau!

Quoted 48 lines

The lines at drivers/usb/host/ehci-hub.c:841
case GetPortStatus:
if (!wIndex || wIndex > ports)
goto error;
wIndex--;
status = 0;
temp = ehci_readl(ehci, status_reg);

are being compiled very differently.

With GCC 4.5.2:
0xc0229810 <ehci_hub_control+672>: cmp r3, #0 ; 0x0 0xc0229814 <ehci_hub_control+676>: beq 0xc0229df8 <ehci_hub_control+2184> 0xc0229818 <ehci_hub_control+680>: cmp r3, r0
0xc022981c <ehci_hub_control+684>: bgt 0xc0229df8 <ehci_hub_control+2184> 0xc0229820 <ehci_hub_control+688>: sub r8, r3, #1 ; 0x1 0xc0229824 <ehci_hub_control+692>: ldr r5, [r7, #4] 0xc0229828 <ehci_hub_control+696>: uxth r8, r8
0xc022982c <ehci_hub_control+700>: dmb sy

With GCC 4.6.0:
0xc0221630 <ehci_hub_control+808>: cmp r3, #0 ; 0x0 0xc0221634 <ehci_hub_control+812>: beq 0xc0221d7c <ehci_hub_control+2676> 0xc0221638 <ehci_hub_control+816>: cmp r3, r0
0xc022163c <ehci_hub_control+820>: bgt 0xc0221d7c <ehci_hub_control+2676> 0xc0221640 <ehci_hub_control+824>: sub r7, r3, #1 ; 0x1 0xc0221644 <ehci_hub_control+828>: add r3, r11, #16 ; 0x10 0xc0221648 <ehci_hub_control+832>: add r3, r8, r3, lsl #2 0xc022164c <ehci_hub_control+836>: uxth r7, r7
0xc0221650 <ehci_hub_control+840>: ldrb r5, [r3, #5] 0xc0221654 <ehci_hub_control+844>: ldrb r2, [r3, #4] 0xc0221658 <ehci_hub_control+848>: orr r5, r2, r5, lsl #8 0xc022165c <ehci_hub_control+852>: ldrb r2, [r3, #6] 0xc0221660 <ehci_hub_control+856>: ldrb r3, [r3, #7] 0xc0221664 <ehci_hub_control+860>: orr r5, r5, r2, lsl #16 0xc0221668 <ehci_hub_control+864>: orr r5, r5, r3, lsl #24 0xc022166c <ehci_hub_control+868>: dmb sy

As you can see, the old compiler accesses the ehci status register in a single access, the new compiler dances around and makes multiple accesses, which in the end get very incorrect data (I think that this register is like many which clear bits on reads).

Any ideas where I can go with this? I'm really trying to keep up with Poky/Yocto, but this move to GCC 4.6.0 has broken my ARM targets :-( I do have PowerPC targets as well - they seem fine (from limited testing) with the new compiler.

Quoted 6 lines

Note: if you want to see the whole function disassembly, look at
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.5.2
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.6.0

Thanks

_______________________________________________
poky mailing list
[email protected]
https://lists.yoctoproject.org/listinfo/poky

Signature · 3 lines
Darren Hart
Intel Open Source Technology Center
Yocto Project - Linux Kernel
#4 lore raw
In reply to Saul Wold
Quoted 19 lines · Saul Wold
Quoted 11 lines · Gary Thomas

Now that Poky is using GCC 4.6.0, has anyone actually checked the operation on the BeagleBoard? I suspect that you'll find that the EHCI USB does not work. I can't really check here as I don't trust the EHCI on my BeagleBoard rev C3 (not xM).

That said, I've tried this new compiler (previously reported) on my own OMAP/3530 board which uses the same 2.6.37 kernel (just not the linux-yocto version). When I build my kernel with GCC 4.5.2, it works perfectly, but fails with GCC 4.6.0 (I trust the hardware).

Quoted 6 lines

I've isolated it down to at least the function 'ehci_hub_control' (but I suspect the problem is more fundamental). Comparing the code generated by the two compilers with the same source tree, this function is dramatically different. I can see why it's failing, I just don't know why the compiler is doing what it's doing.

Quoted 50 lines

Gary,

I am not sure if Khem Raj is on this list, so I am forwarding it to him, he might have some insight.

Sau!

Quoted 42 lines

The lines at drivers/usb/host/ehci-hub.c:841
case GetPortStatus:
if (!wIndex || wIndex > ports)
goto error;
wIndex--;
status = 0;
temp = ehci_readl(ehci, status_reg);

are being compiled very differently.

With GCC 4.5.2:
0xc0229810 <ehci_hub_control+672>: cmp r3, #0 ; 0x0 0xc0229814 <ehci_hub_control+676>: beq 0xc0229df8 <ehci_hub_control+2184> 0xc0229818 <ehci_hub_control+680>: cmp r3, r0
0xc022981c <ehci_hub_control+684>: bgt 0xc0229df8 <ehci_hub_control+2184> 0xc0229820 <ehci_hub_control+688>: sub r8, r3, #1 ; 0x1 0xc0229824 <ehci_hub_control+692>: ldr r5, [r7, #4] 0xc0229828 <ehci_hub_control+696>: uxth r8, r8
0xc022982c <ehci_hub_control+700>: dmb sy

With GCC 4.6.0:
0xc0221630 <ehci_hub_control+808>: cmp r3, #0 ; 0x0 0xc0221634 <ehci_hub_control+812>: beq 0xc0221d7c <ehci_hub_control+2676> 0xc0221638 <ehci_hub_control+816>: cmp r3, r0
0xc022163c <ehci_hub_control+820>: bgt 0xc0221d7c <ehci_hub_control+2676> 0xc0221640 <ehci_hub_control+824>: sub r7, r3, #1 ; 0x1 0xc0221644 <ehci_hub_control+828>: add r3, r11, #16 ; 0x10 0xc0221648 <ehci_hub_control+832>: add r3, r8, r3, lsl #2 0xc022164c <ehci_hub_control+836>: uxth r7, r7
0xc0221650 <ehci_hub_control+840>: ldrb r5, [r3, #5] 0xc0221654 <ehci_hub_control+844>: ldrb r2, [r3, #4] 0xc0221658 <ehci_hub_control+848>: orr r5, r2, r5, lsl #8 0xc022165c <ehci_hub_control+852>: ldrb r2, [r3, #6] 0xc0221660 <ehci_hub_control+856>: ldrb r3, [r3, #7] 0xc0221664 <ehci_hub_control+860>: orr r5, r5, r2, lsl #16 0xc0221668 <ehci_hub_control+864>: orr r5, r5, r3, lsl #24 0xc022166c <ehci_hub_control+868>: dmb sy

As you can see, the old compiler accesses the ehci status register in a single access, the new compiler dances around and makes multiple accesses, which in the end get very incorrect data (I think that this register is like many which clear bits on reads).

ok seems like unaligned loads are kicking in may be its missing some qualifier. Can you do me a favour ? Please send me preprocessed output of the source file and let me see what I find

-Khem

#3 lore raw
In reply to Darren Hart
Quoted 23 lines · Darren Hart
Quoted 20 lines · Saul Wold
Quoted 10 lines · Gary Thomas

Now that Poky is using GCC 4.6.0, has anyone actually checked the operation on the BeagleBoard? I suspect that you'll find that the EHCI USB does not work. I can't really check here as I don't trust the EHCI on my BeagleBoard rev C3 (not xM).

That said, I've tried this new compiler (previously reported) on my own OMAP/3530 board which uses the same 2.6.37 kernel (just not the linux-yocto version). When I build my kernel with GCC 4.5.2, it works perfectly, but fails with GCC 4.6.0 (I trust the hardware).

Quoted 6 lines

I've isolated it down to at least the function 'ehci_hub_control' (but I suspect the problem is more fundamental). Comparing the code generated by the two compilers with the same source tree, this function is dramatically different. I can see why it's failing, I just don't know why the compiler is doing what it's doing.

Gary,

Quoted 7 lines
Quoted 2 lines

I am not sure if Khem Raj is on this list, so I am forwarding it to him, he might have some insight.

By switching GCCVERSION to 4.5.1, a kernel that previously failed to bring up the USB hub on the beagleboard xM (and the ethernet port) with 4.6.0, now successfully does so. Unfortunatley, the framebuffer device fails now.

Turns out this was user error. In my development kernel I now have usb,ethernet,audio, and the sato desktop all running when building the kernel with gcc 4.5.1.

I'll send a more detailed beagleboard status to the list as several folks seem to be interested.

--
Darren

Quoted 69 lines

Any hints from the compiler crowd on things to try? Or output we can collect to help resolve this?

--
Darren

Quoted 62 lines

Sau!

Quoted 48 lines

The lines at drivers/usb/host/ehci-hub.c:841
case GetPortStatus:
if (!wIndex || wIndex > ports)
goto error;
wIndex--;
status = 0;
temp = ehci_readl(ehci, status_reg);

are being compiled very differently.

With GCC 4.5.2:
0xc0229810 <ehci_hub_control+672>: cmp r3, #0 ; 0x0 0xc0229814 <ehci_hub_control+676>: beq 0xc0229df8 <ehci_hub_control+2184> 0xc0229818 <ehci_hub_control+680>: cmp r3, r0
0xc022981c <ehci_hub_control+684>: bgt 0xc0229df8 <ehci_hub_control+2184> 0xc0229820 <ehci_hub_control+688>: sub r8, r3, #1 ; 0x1 0xc0229824 <ehci_hub_control+692>: ldr r5, [r7, #4] 0xc0229828 <ehci_hub_control+696>: uxth r8, r8
0xc022982c <ehci_hub_control+700>: dmb sy

With GCC 4.6.0:
0xc0221630 <ehci_hub_control+808>: cmp r3, #0 ; 0x0 0xc0221634 <ehci_hub_control+812>: beq 0xc0221d7c <ehci_hub_control+2676> 0xc0221638 <ehci_hub_control+816>: cmp r3, r0
0xc022163c <ehci_hub_control+820>: bgt 0xc0221d7c <ehci_hub_control+2676> 0xc0221640 <ehci_hub_control+824>: sub r7, r3, #1 ; 0x1 0xc0221644 <ehci_hub_control+828>: add r3, r11, #16 ; 0x10 0xc0221648 <ehci_hub_control+832>: add r3, r8, r3, lsl #2 0xc022164c <ehci_hub_control+836>: uxth r7, r7
0xc0221650 <ehci_hub_control+840>: ldrb r5, [r3, #5] 0xc0221654 <ehci_hub_control+844>: ldrb r2, [r3, #4] 0xc0221658 <ehci_hub_control+848>: orr r5, r2, r5, lsl #8 0xc022165c <ehci_hub_control+852>: ldrb r2, [r3, #6] 0xc0221660 <ehci_hub_control+856>: ldrb r3, [r3, #7] 0xc0221664 <ehci_hub_control+860>: orr r5, r5, r2, lsl #16 0xc0221668 <ehci_hub_control+864>: orr r5, r5, r3, lsl #24 0xc022166c <ehci_hub_control+868>: dmb sy

As you can see, the old compiler accesses the ehci status register in a single access, the new compiler dances around and makes multiple accesses, which in the end get very incorrect data (I think that this register is like many which clear bits on reads).

Any ideas where I can go with this? I'm really trying to keep up with Poky/Yocto, but this move to GCC 4.6.0 has broken my ARM targets :-( I do have PowerPC targets as well - they seem fine (from limited testing) with the new compiler.

Quoted 6 lines

Note: if you want to see the whole function disassembly, look at
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.5.2
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.6.0

Thanks

_______________________________________________
poky mailing list
[email protected]
https://lists.yoctoproject.org/listinfo/poky

Signature · 3 lines
Darren Hart
Intel Open Source Technology Center
Yocto Project - Linux Kernel
#5 lore raw
In reply to Saul Wold
Quoted 18 lines · Saul Wold
Quoted 10 lines · Gary Thomas

Now that Poky is using GCC 4.6.0, has anyone actually checked the operation on the BeagleBoard? I suspect that you'll find that the EHCI USB does not work. I can't really check here as I don't trust the EHCI on my BeagleBoard rev C3 (not xM).

That said, I've tried this new compiler (previously reported) on my own OMAP/3530 board which uses the same 2.6.37 kernel (just not the linux-yocto version). When I build my kernel with GCC 4.5.2, it works perfectly, but fails with GCC 4.6.0 (I trust the hardware).

Quoted 6 lines

I've isolated it down to at least the function 'ehci_hub_control' (but I suspect the problem is more fundamental). Comparing the code generated by the two compilers with the same source tree, this function is dramatically different. I can see why it's failing, I just don't know why the compiler is doing what it's doing.

Quoted 62 lines

Gary,

I am not sure if Khem Raj is on this list, so I am forwarding it to him, he might have some insight.

Sau!

Quoted 54 lines

The lines at drivers/usb/host/ehci-hub.c:841
case GetPortStatus:
if (!wIndex || wIndex > ports)
goto error;
wIndex--;
status = 0;
temp = ehci_readl(ehci, status_reg);

are being compiled very differently.

With GCC 4.5.2:
0xc0229810 <ehci_hub_control+672>: cmp r3, #0 ; 0x0 0xc0229814 <ehci_hub_control+676>: beq 0xc0229df8 <ehci_hub_control+2184> 0xc0229818 <ehci_hub_control+680>: cmp r3, r0
0xc022981c <ehci_hub_control+684>: bgt 0xc0229df8 <ehci_hub_control+2184> 0xc0229820 <ehci_hub_control+688>: sub r8, r3, #1 ; 0x1 0xc0229824 <ehci_hub_control+692>: ldr r5, [r7, #4] 0xc0229828 <ehci_hub_control+696>: uxth r8, r8
0xc022982c <ehci_hub_control+700>: dmb sy

With GCC 4.6.0:
0xc0221630 <ehci_hub_control+808>: cmp r3, #0 ; 0x0 0xc0221634 <ehci_hub_control+812>: beq 0xc0221d7c <ehci_hub_control+2676> 0xc0221638 <ehci_hub_control+816>: cmp r3, r0
0xc022163c <ehci_hub_control+820>: bgt 0xc0221d7c <ehci_hub_control+2676> 0xc0221640 <ehci_hub_control+824>: sub r7, r3, #1 ; 0x1 0xc0221644 <ehci_hub_control+828>: add r3, r11, #16 ; 0x10 0xc0221648 <ehci_hub_control+832>: add r3, r8, r3, lsl #2 0xc022164c <ehci_hub_control+836>: uxth r7, r7
0xc0221650 <ehci_hub_control+840>: ldrb r5, [r3, #5] 0xc0221654 <ehci_hub_control+844>: ldrb r2, [r3, #4] 0xc0221658 <ehci_hub_control+848>: orr r5, r2, r5, lsl #8 0xc022165c <ehci_hub_control+852>: ldrb r2, [r3, #6] 0xc0221660 <ehci_hub_control+856>: ldrb r3, [r3, #7] 0xc0221664 <ehci_hub_control+860>: orr r5, r5, r2, lsl #16 0xc0221668 <ehci_hub_control+864>: orr r5, r5, r3, lsl #24 0xc022166c <ehci_hub_control+868>: dmb sy

As you can see, the old compiler accesses the ehci status register in a single access, the new compiler dances around and makes multiple accesses, which in the end get very incorrect data (I think that this register is like many which clear bits on reads).

Any ideas where I can go with this? I'm really trying to keep up with Poky/Yocto, but this move to GCC 4.6.0 has broken my ARM targets :-( I do have PowerPC targets as well - they seem fine (from limited testing) with the new compiler.

Note: if you want to see the whole function disassembly, look at
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.5.2
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.6.0

Thanks

OK. thanks for sending the sample code.
Can you add -fstrict-volatile-bitfields option to CFLAGS while compiling this file or even the kernel for test sake ? and see if the problem goes away ?

-Khem

#6 lore raw
Quoted 79 lines · Khem Raj
Quoted 78 lines · Saul Wold
Quoted 10 lines · Gary Thomas

Now that Poky is using GCC 4.6.0, has anyone actually checked the operation on the BeagleBoard? I suspect that you'll find that the EHCI USB does not work. I can't really check here as I don't trust the EHCI on my BeagleBoard rev C3 (not xM).

That said, I've tried this new compiler (previously reported) on my own OMAP/3530 board which uses the same 2.6.37 kernel (just not the linux-yocto version). When I build my kernel with GCC 4.5.2, it works perfectly, but fails with GCC 4.6.0 (I trust the hardware).

Quoted 6 lines

I've isolated it down to at least the function 'ehci_hub_control' (but I suspect the problem is more fundamental). Comparing the code generated by the two compilers with the same source tree, this function is dramatically different. I can see why it's failing, I just don't know why the compiler is doing what it's doing.

Gary,

I am not sure if Khem Raj is on this list, so I am forwarding it to him, he might have some insight.

Sau!

Quoted 52 lines

The lines at drivers/usb/host/ehci-hub.c:841
case GetPortStatus:
if (!wIndex || wIndex > ports)
goto error;
wIndex--;
status = 0;
temp = ehci_readl(ehci, status_reg);

are being compiled very differently.

With GCC 4.5.2:
0xc0229810 <ehci_hub_control+672>: cmp r3, #0 ; 0x0 0xc0229814 <ehci_hub_control+676>: beq 0xc0229df8 <ehci_hub_control+2184> 0xc0229818 <ehci_hub_control+680>: cmp r3, r0
0xc022981c <ehci_hub_control+684>: bgt 0xc0229df8 <ehci_hub_control+2184> 0xc0229820 <ehci_hub_control+688>: sub r8, r3, #1 ; 0x1 0xc0229824 <ehci_hub_control+692>: ldr r5, [r7, #4] 0xc0229828 <ehci_hub_control+696>: uxth r8, r8
0xc022982c <ehci_hub_control+700>: dmb sy

With GCC 4.6.0:
0xc0221630 <ehci_hub_control+808>: cmp r3, #0 ; 0x0 0xc0221634 <ehci_hub_control+812>: beq 0xc0221d7c <ehci_hub_control+2676> 0xc0221638 <ehci_hub_control+816>: cmp r3, r0
0xc022163c <ehci_hub_control+820>: bgt 0xc0221d7c <ehci_hub_control+2676> 0xc0221640 <ehci_hub_control+824>: sub r7, r3, #1 ; 0x1 0xc0221644 <ehci_hub_control+828>: add r3, r11, #16 ; 0x10 0xc0221648 <ehci_hub_control+832>: add r3, r8, r3, lsl #2 0xc022164c <ehci_hub_control+836>: uxth r7, r7
0xc0221650 <ehci_hub_control+840>: ldrb r5, [r3, #5] 0xc0221654 <ehci_hub_control+844>: ldrb r2, [r3, #4] 0xc0221658 <ehci_hub_control+848>: orr r5, r2, r5, lsl #8 0xc022165c <ehci_hub_control+852>: ldrb r2, [r3, #6] 0xc0221660 <ehci_hub_control+856>: ldrb r3, [r3, #7] 0xc0221664 <ehci_hub_control+860>: orr r5, r5, r2, lsl #16 0xc0221668 <ehci_hub_control+864>: orr r5, r5, r3, lsl #24 0xc022166c <ehci_hub_control+868>: dmb sy

As you can see, the old compiler accesses the ehci status register in a single access, the new compiler dances around and makes multiple accesses, which in the end get very incorrect data (I think that this register is like many which clear bits on reads).

Any ideas where I can go with this? I'm really trying to keep up with Poky/Yocto, but this move to GCC 4.6.0 has broken my ARM targets :-( I do have PowerPC targets as well - they seem fine (from limited testing) with the new compiler.

Note: if you want to see the whole function disassembly, look at
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.5.2
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.6.0

Quoted 6 lines
Quoted 2 lines
Quoted 2 lines

Thanks

OK. thanks for sending the sample code.
Can you add -fstrict-volatile-bitfields option to CFLAGS while compiling this file or even the kernel for test sake ? and see if the problem goes away ?

Sorry, this doesn't seem to make any difference in the generated code. Hardly surprising as the register in question is a simple u32, not a bitfield.

Signature · 4 lines
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------
#7 lore raw
Quoted 88 lines · Gary Thomas
Quoted 87 lines · Khem Raj
Quoted 85 lines · Saul Wold
Quoted 10 lines · Gary Thomas

Now that Poky is using GCC 4.6.0, has anyone actually checked the operation on the BeagleBoard? I suspect that you'll find that the EHCI USB does not work. I can't really check here as I don't trust the EHCI on my BeagleBoard rev C3 (not xM).

That said, I've tried this new compiler (previously reported) on my own OMAP/3530 board which uses the same 2.6.37 kernel (just not the linux-yocto version). When I build my kernel with GCC 4.5.2, it works perfectly, but fails with GCC 4.6.0 (I trust the hardware).

Quoted 6 lines

I've isolated it down to at least the function 'ehci_hub_control' (but I suspect the problem is more fundamental). Comparing the code generated by the two compilers with the same source tree, this function is dramatically different. I can see why it's failing, I just don't know why the compiler is doing what it's doing.

Gary,

I am not sure if Khem Raj is on this list, so I am forwarding it to him, he might have some insight.

Sau!

Quoted 59 lines

The lines at drivers/usb/host/ehci-hub.c:841
case GetPortStatus:
if (!wIndex || wIndex > ports)
goto error;
wIndex--;
status = 0;
temp = ehci_readl(ehci, status_reg);

are being compiled very differently.

With GCC 4.5.2:
0xc0229810 <ehci_hub_control+672>: cmp r3, #0 ; 0x0 0xc0229814 <ehci_hub_control+676>: beq 0xc0229df8 <ehci_hub_control+2184>
0xc0229818 <ehci_hub_control+680>: cmp r3, r0 0xc022981c <ehci_hub_control+684>: bgt 0xc0229df8 <ehci_hub_control+2184>
0xc0229820 <ehci_hub_control+688>: sub r8, r3, #1 ; 0x1 0xc0229824 <ehci_hub_control+692>: ldr r5, [r7, #4] 0xc0229828 <ehci_hub_control+696>: uxth r8, r8 0xc022982c <ehci_hub_control+700>: dmb sy

With GCC 4.6.0:
0xc0221630 <ehci_hub_control+808>: cmp r3, #0 ; 0x0 0xc0221634 <ehci_hub_control+812>: beq 0xc0221d7c <ehci_hub_control+2676>
0xc0221638 <ehci_hub_control+816>: cmp r3, r0 0xc022163c <ehci_hub_control+820>: bgt 0xc0221d7c <ehci_hub_control+2676>
0xc0221640 <ehci_hub_control+824>: sub r7, r3, #1 ; 0x1 0xc0221644 <ehci_hub_control+828>: add r3, r11, #16 ; 0x10 0xc0221648 <ehci_hub_control+832>: add r3, r8, r3, lsl #2 0xc022164c <ehci_hub_control+836>: uxth r7, r7 0xc0221650 <ehci_hub_control+840>: ldrb r5, [r3, #5] 0xc0221654 <ehci_hub_control+844>: ldrb r2, [r3, #4] 0xc0221658 <ehci_hub_control+848>: orr r5, r2, r5, lsl #8 0xc022165c <ehci_hub_control+852>: ldrb r2, [r3, #6] 0xc0221660 <ehci_hub_control+856>: ldrb r3, [r3, #7] 0xc0221664 <ehci_hub_control+860>: orr r5, r5, r2, lsl #16 0xc0221668 <ehci_hub_control+864>: orr r5, r5, r3, lsl #24 0xc022166c <ehci_hub_control+868>: dmb sy

As you can see, the old compiler accesses the ehci status register in a single access, the new compiler dances around and makes multiple accesses, which in the end get very incorrect data (I think that this register is like many which clear bits on reads).

Any ideas where I can go with this? I'm really trying to keep up with Poky/Yocto, but this move to GCC 4.6.0 has broken my ARM targets :-( I do have PowerPC targets as well - they seem fine (from limited testing)
with the new compiler.

Note: if you want to see the whole function disassembly, look at
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.5.2
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.6.0

Thanks

Quoted 9 lines
Quoted 4 lines

OK. thanks for sending the sample code.
Can you add -fstrict-volatile-bitfields option to CFLAGS while compiling this file or even the kernel for test sake ? and see if the problem goes away ?

Sorry, this doesn't seem to make any difference in the generated code. Hardly surprising as the register in question is a simple u32, not a bitfield.

No. Its not a simple u32
its a pointer and points to address of a member of structure type ehci_hcd which can have bitfields. There has been bug reports in this area in GCC which have been fixed and some are not. I was merely trying to reduce the problem. So you tried compiling kernel with this option and you found that it does not work ? Did I understand it correct

#8 lore raw
In reply to Gary Thomas
Quoted 88 lines · Gary Thomas
Quoted 87 lines · Khem Raj
Quoted 85 lines · Saul Wold
Quoted 10 lines · Gary Thomas

Now that Poky is using GCC 4.6.0, has anyone actually checked the operation on the BeagleBoard? I suspect that you'll find that the EHCI USB does not work. I can't really check here as I don't trust the EHCI on my BeagleBoard rev C3 (not xM).

That said, I've tried this new compiler (previously reported) on my own OMAP/3530 board which uses the same 2.6.37 kernel (just not the linux-yocto version). When I build my kernel with GCC 4.5.2, it works perfectly, but fails with GCC 4.6.0 (I trust the hardware).

Quoted 6 lines

I've isolated it down to at least the function 'ehci_hub_control' (but I suspect the problem is more fundamental). Comparing the code generated by the two compilers with the same source tree, this function is dramatically different. I can see why it's failing, I just don't know why the compiler is doing what it's doing.

Gary,

I am not sure if Khem Raj is on this list, so I am forwarding it to him, he might have some insight.

Sau!

Quoted 59 lines

The lines at drivers/usb/host/ehci-hub.c:841
case GetPortStatus:
if (!wIndex || wIndex > ports)
goto error;
wIndex--;
status = 0;
temp = ehci_readl(ehci, status_reg);

are being compiled very differently.

With GCC 4.5.2:
0xc0229810 <ehci_hub_control+672>: cmp r3, #0 ; 0x0 0xc0229814 <ehci_hub_control+676>: beq 0xc0229df8 <ehci_hub_control+2184>
0xc0229818 <ehci_hub_control+680>: cmp r3, r0 0xc022981c <ehci_hub_control+684>: bgt 0xc0229df8 <ehci_hub_control+2184>
0xc0229820 <ehci_hub_control+688>: sub r8, r3, #1 ; 0x1 0xc0229824 <ehci_hub_control+692>: ldr r5, [r7, #4] 0xc0229828 <ehci_hub_control+696>: uxth r8, r8 0xc022982c <ehci_hub_control+700>: dmb sy

With GCC 4.6.0:
0xc0221630 <ehci_hub_control+808>: cmp r3, #0 ; 0x0 0xc0221634 <ehci_hub_control+812>: beq 0xc0221d7c <ehci_hub_control+2676>
0xc0221638 <ehci_hub_control+816>: cmp r3, r0 0xc022163c <ehci_hub_control+820>: bgt 0xc0221d7c <ehci_hub_control+2676>
0xc0221640 <ehci_hub_control+824>: sub r7, r3, #1 ; 0x1 0xc0221644 <ehci_hub_control+828>: add r3, r11, #16 ; 0x10 0xc0221648 <ehci_hub_control+832>: add r3, r8, r3, lsl #2 0xc022164c <ehci_hub_control+836>: uxth r7, r7 0xc0221650 <ehci_hub_control+840>: ldrb r5, [r3, #5] 0xc0221654 <ehci_hub_control+844>: ldrb r2, [r3, #4] 0xc0221658 <ehci_hub_control+848>: orr r5, r2, r5, lsl #8 0xc022165c <ehci_hub_control+852>: ldrb r2, [r3, #6] 0xc0221660 <ehci_hub_control+856>: ldrb r3, [r3, #7] 0xc0221664 <ehci_hub_control+860>: orr r5, r5, r2, lsl #16 0xc0221668 <ehci_hub_control+864>: orr r5, r5, r3, lsl #24 0xc022166c <ehci_hub_control+868>: dmb sy

As you can see, the old compiler accesses the ehci status register in a single access, the new compiler dances around and makes multiple accesses, which in the end get very incorrect data (I think that this register is like many which clear bits on reads).

Any ideas where I can go with this? I'm really trying to keep up with Poky/Yocto, but this move to GCC 4.6.0 has broken my ARM targets :-( I do have PowerPC targets as well - they seem fine (from limited testing)
with the new compiler.

Note: if you want to see the whole function disassembly, look at
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.5.2
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.6.0

Thanks

Quoted 9 lines
Quoted 4 lines

OK. thanks for sending the sample code.
Can you add -fstrict-volatile-bitfields option to CFLAGS while compiling this file or even the kernel for test sake ? and see if the problem goes away ?

Sorry, this doesn't seem to make any difference in the generated code. Hardly surprising as the register in question is a simple u32, not a bitfield.

I reduced the testcase a bit further and it seems the bug is similar to http://gcc.gnu.org/bugzilla/PR45819

but a bit different that in this case -fstrict-volatile-bitfields does not help

a fix for this particular problem is

include/linux/usb/ehci_def.h +1 -1
index 6563802..b8c1833 100644
@@ -194,7 +194,7 @@ struct ehci_dbg_port {
        u32     data47;
        u32     address;
 #define DBGP_EPADDR(dev, ep)   (((dev)<<8)|(ep))
-} __attribute__ ((packed));
+} __attribute__ ((packed,aligned(__alignof__(int))));
 
 #ifdef CONFIG_EARLY_PRINTK_DBGP
 #include <linux/init.h>

since this particular structure all elements are integers but there might be more packed structures which have say char or short ints and are not necessarily ordered in the descending order of alignment

Try it out and see if it helps

#9 lore raw
Quoted 120 lines · Khem Raj
Quoted 88 lines · Gary Thomas
Quoted 87 lines · Khem Raj
Quoted 85 lines · Saul Wold
Quoted 10 lines · Gary Thomas

Now that Poky is using GCC 4.6.0, has anyone actually checked the operation on the BeagleBoard? I suspect that you'll find that the EHCI USB does not work. I can't really check here as I don't trust the EHCI on my BeagleBoard rev C3 (not xM).

That said, I've tried this new compiler (previously reported) on my own OMAP/3530 board which uses the same 2.6.37 kernel (just not the linux-yocto version). When I build my kernel with GCC 4.5.2, it works perfectly, but fails with GCC 4.6.0 (I trust the hardware).

Quoted 6 lines

I've isolated it down to at least the function 'ehci_hub_control' (but I suspect the problem is more fundamental). Comparing the code generated by the two compilers with the same source tree, this function is dramatically different. I can see why it's failing, I just don't know why the compiler is doing what it's doing.

Gary,

I am not sure if Khem Raj is on this list, so I am forwarding it to him, he might have some insight.

Sau!

Quoted 59 lines

The lines at drivers/usb/host/ehci-hub.c:841
case GetPortStatus:
if (!wIndex || wIndex > ports)
goto error;
wIndex--;
status = 0;
temp = ehci_readl(ehci, status_reg);

are being compiled very differently.

With GCC 4.5.2:
0xc0229810 <ehci_hub_control+672>: cmp r3, #0 ; 0x0 0xc0229814 <ehci_hub_control+676>: beq 0xc0229df8 <ehci_hub_control+2184>
0xc0229818 <ehci_hub_control+680>: cmp r3, r0 0xc022981c <ehci_hub_control+684>: bgt 0xc0229df8 <ehci_hub_control+2184>
0xc0229820 <ehci_hub_control+688>: sub r8, r3, #1 ; 0x1 0xc0229824 <ehci_hub_control+692>: ldr r5, [r7, #4] 0xc0229828 <ehci_hub_control+696>: uxth r8, r8 0xc022982c <ehci_hub_control+700>: dmb sy

With GCC 4.6.0:
0xc0221630 <ehci_hub_control+808>: cmp r3, #0 ; 0x0 0xc0221634 <ehci_hub_control+812>: beq 0xc0221d7c <ehci_hub_control+2676>
0xc0221638 <ehci_hub_control+816>: cmp r3, r0 0xc022163c <ehci_hub_control+820>: bgt 0xc0221d7c <ehci_hub_control+2676>
0xc0221640 <ehci_hub_control+824>: sub r7, r3, #1 ; 0x1 0xc0221644 <ehci_hub_control+828>: add r3, r11, #16 ; 0x10 0xc0221648 <ehci_hub_control+832>: add r3, r8, r3, lsl #2 0xc022164c <ehci_hub_control+836>: uxth r7, r7 0xc0221650 <ehci_hub_control+840>: ldrb r5, [r3, #5] 0xc0221654 <ehci_hub_control+844>: ldrb r2, [r3, #4] 0xc0221658 <ehci_hub_control+848>: orr r5, r2, r5, lsl #8 0xc022165c <ehci_hub_control+852>: ldrb r2, [r3, #6] 0xc0221660 <ehci_hub_control+856>: ldrb r3, [r3, #7] 0xc0221664 <ehci_hub_control+860>: orr r5, r5, r2, lsl #16 0xc0221668 <ehci_hub_control+864>: orr r5, r5, r3, lsl #24 0xc022166c <ehci_hub_control+868>: dmb sy

As you can see, the old compiler accesses the ehci status register in a single access, the new compiler dances around and makes multiple accesses, which in the end get very incorrect data (I think that this register is like many which clear bits on reads).

Any ideas where I can go with this? I'm really trying to keep up with Poky/Yocto, but this move to GCC 4.6.0 has broken my ARM targets :-( I do have PowerPC targets as well - they seem fine (from limited testing)
with the new compiler.

Note: if you want to see the whole function disassembly, look at
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.5.2
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.6.0

Thanks

Quoted 9 lines
Quoted 4 lines

OK. thanks for sending the sample code.
Can you add -fstrict-volatile-bitfields option to CFLAGS while compiling this file or even the kernel for test sake ? and see if the problem goes away ?

Sorry, this doesn't seem to make any difference in the generated code. Hardly surprising as the register in question is a simple u32, not a bitfield.

I reduced the testcase a bit further and it seems the bug is similar to http://gcc.gnu.org/bugzilla/PR45819

but a bit different that in this case -fstrict-volatile-bitfields does not help

a fix for this particular problem is

include/linux/usb/ehci_def.h +1 -1
index 6563802..b8c1833 100644
@@ -194,7 +194,7 @@ struct ehci_dbg_port {
 u32 data47;
 u32 address;
 #define DBGP_EPADDR(dev, ep) (((dev)<<8)|(ep))
-} __attribute__ ((packed));
+} __attribute__ ((packed,aligned(__alignof__(int))));
 
 #ifdef CONFIG_EARLY_PRINTK_DBGP
 #include <linux/init.h>
Quoted 4 lines

since this particular structure all elements are integers but there might be more packed structures which have say char or short ints and are not necessarily ordered in the descending order of alignment

Try it out and see if it helps

Yes, this does fix the problem :-) The EHCI USB port on the OMAP is working again.

Thanks

Signature · 4 lines
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------
#10 lore raw
Quoted 121 lines · Gary Thomas
Quoted 120 lines · Khem Raj
Quoted 88 lines · Gary Thomas
Quoted 87 lines · Khem Raj
Quoted 85 lines · Saul Wold
Quoted 10 lines · Gary Thomas

Now that Poky is using GCC 4.6.0, has anyone actually checked the operation on the BeagleBoard? I suspect that you'll find that the EHCI USB does not work. I can't really check here as I don't trust the EHCI on my BeagleBoard rev C3 (not xM).

That said, I've tried this new compiler (previously reported) on my own OMAP/3530 board which uses the same 2.6.37 kernel (just not the linux-yocto version). When I build my kernel with GCC 4.5.2, it works perfectly, but fails with GCC 4.6.0 (I trust the hardware).

Quoted 6 lines

I've isolated it down to at least the function 'ehci_hub_control' (but I suspect the problem is more fundamental). Comparing the code generated by the two compilers with the same source tree, this function is dramatically different. I can see why it's failing, I just don't know why the compiler is doing what it's doing.

Gary,

I am not sure if Khem Raj is on this list, so I am forwarding it to him, he might have some insight.

Sau!

Quoted 59 lines

The lines at drivers/usb/host/ehci-hub.c:841
case GetPortStatus:
if (!wIndex || wIndex > ports)
goto error;
wIndex--;
status = 0;
temp = ehci_readl(ehci, status_reg);

are being compiled very differently.

With GCC 4.5.2:
0xc0229810 <ehci_hub_control+672>: cmp r3, #0 ; 0x0 0xc0229814 <ehci_hub_control+676>: beq 0xc0229df8 <ehci_hub_control+2184>
0xc0229818 <ehci_hub_control+680>: cmp r3, r0 0xc022981c <ehci_hub_control+684>: bgt 0xc0229df8 <ehci_hub_control+2184>
0xc0229820 <ehci_hub_control+688>: sub r8, r3, #1 ; 0x1 0xc0229824 <ehci_hub_control+692>: ldr r5, [r7, #4] 0xc0229828 <ehci_hub_control+696>: uxth r8, r8 0xc022982c <ehci_hub_control+700>: dmb sy

With GCC 4.6.0:
0xc0221630 <ehci_hub_control+808>: cmp r3, #0 ; 0x0 0xc0221634 <ehci_hub_control+812>: beq 0xc0221d7c <ehci_hub_control+2676>
0xc0221638 <ehci_hub_control+816>: cmp r3, r0 0xc022163c <ehci_hub_control+820>: bgt 0xc0221d7c <ehci_hub_control+2676>
0xc0221640 <ehci_hub_control+824>: sub r7, r3, #1 ; 0x1 0xc0221644 <ehci_hub_control+828>: add r3, r11, #16 ; 0x10 0xc0221648 <ehci_hub_control+832>: add r3, r8, r3, lsl #2 0xc022164c <ehci_hub_control+836>: uxth r7, r7 0xc0221650 <ehci_hub_control+840>: ldrb r5, [r3, #5] 0xc0221654 <ehci_hub_control+844>: ldrb r2, [r3, #4] 0xc0221658 <ehci_hub_control+848>: orr r5, r2, r5, lsl #8 0xc022165c <ehci_hub_control+852>: ldrb r2, [r3, #6] 0xc0221660 <ehci_hub_control+856>: ldrb r3, [r3, #7] 0xc0221664 <ehci_hub_control+860>: orr r5, r5, r2, lsl #16 0xc0221668 <ehci_hub_control+864>: orr r5, r5, r3, lsl #24 0xc022166c <ehci_hub_control+868>: dmb sy

As you can see, the old compiler accesses the ehci status register in a single access, the new compiler dances around and makes multiple accesses, which in the end get very incorrect data (I think that this register is like many which clear bits on reads).

Any ideas where I can go with this? I'm really trying to keep up with Poky/Yocto, but this move to GCC 4.6.0 has broken my ARM targets :-( I do have PowerPC targets as well - they seem fine (from limited testing)
with the new compiler.

Note: if you want to see the whole function disassembly, look at
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.5.2
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.6.0

Thanks

Quoted 9 lines
Quoted 4 lines

OK. thanks for sending the sample code.
Can you add -fstrict-volatile-bitfields option to CFLAGS while compiling this file or even the kernel for test sake ? and see if the problem goes away ?

Sorry, this doesn't seem to make any difference in the generated code. Hardly surprising as the register in question is a simple u32, not a bitfield.

I reduced the testcase a bit further and it seems the bug is similar to http://gcc.gnu.org/bugzilla/PR45819

but a bit different that in this case -fstrict-volatile-bitfields does not help

a fix for this particular problem is

include/linux/usb/ehci_def.h +1 -1
index 6563802..b8c1833 100644
@@ -194,7 +194,7 @@ struct ehci_dbg_port {
 u32 data47;
 u32 address;
 #define DBGP_EPADDR(dev, ep) (((dev)<<8)|(ep))
-} __attribute__ ((packed));
+} __attribute__ ((packed,aligned(__alignof__(int))));
 
 #ifdef CONFIG_EARLY_PRINTK_DBGP
 #include <linux/init.h>
Quoted 7 lines
Quoted 4 lines

since this particular structure all elements are integers but there might be more packed structures which have say char or short ints and are not necessarily ordered in the descending order of alignment

Try it out and see if it helps

Yes, this does fix the problem :-) The EHCI USB port on the OMAP is working again.

Note: it also works properly if the __attribute__ ((packed)) is simply removed. Looking at the layout of this structure, I don't see what that attribute is trying to accomplish - every item within the structure is u32[], so IMO there is nothing to pack. Perhaps this observation can help the GCC folks figure out why things are getting confused.

Thanks again

Signature · 4 lines
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------
#11 lore raw
Quoted 130 lines · Gary Thomas
Quoted 121 lines · Gary Thomas
Quoted 120 lines · Khem Raj
Quoted 88 lines · Gary Thomas
Quoted 87 lines · Khem Raj
Quoted 85 lines · Saul Wold
Quoted 10 lines · Gary Thomas

Now that Poky is using GCC 4.6.0, has anyone actually checked the operation on the BeagleBoard? I suspect that you'll find that the EHCI USB does not work. I can't really check here as I don't trust the EHCI on my BeagleBoard rev C3 (not xM).

That said, I've tried this new compiler (previously reported) on my own OMAP/3530 board which uses the same 2.6.37 kernel (just not the linux-yocto version). When I build my kernel with GCC 4.5.2, it works perfectly, but fails with GCC 4.6.0 (I trust the hardware).

Quoted 6 lines

I've isolated it down to at least the function 'ehci_hub_control' (but I suspect the problem is more fundamental). Comparing the code generated by the two compilers with the same source tree, this function is dramatically different. I can see why it's failing, I just don't know why the compiler is doing what it's doing.

Gary,

I am not sure if Khem Raj is on this list, so I am forwarding it to him, he might have some insight.

Sau!

Quoted 59 lines

The lines at drivers/usb/host/ehci-hub.c:841
case GetPortStatus:
if (!wIndex || wIndex > ports)
goto error;
wIndex--;
status = 0;
temp = ehci_readl(ehci, status_reg);

are being compiled very differently.

With GCC 4.5.2:
0xc0229810 <ehci_hub_control+672>: cmp r3, #0 ; 0x0 0xc0229814 <ehci_hub_control+676>: beq 0xc0229df8 <ehci_hub_control+2184>
0xc0229818 <ehci_hub_control+680>: cmp r3, r0 0xc022981c <ehci_hub_control+684>: bgt 0xc0229df8 <ehci_hub_control+2184>
0xc0229820 <ehci_hub_control+688>: sub r8, r3, #1 ; 0x1 0xc0229824 <ehci_hub_control+692>: ldr r5, [r7, #4] 0xc0229828 <ehci_hub_control+696>: uxth r8, r8 0xc022982c <ehci_hub_control+700>: dmb sy

With GCC 4.6.0:
0xc0221630 <ehci_hub_control+808>: cmp r3, #0 ; 0x0 0xc0221634 <ehci_hub_control+812>: beq 0xc0221d7c <ehci_hub_control+2676>
0xc0221638 <ehci_hub_control+816>: cmp r3, r0 0xc022163c <ehci_hub_control+820>: bgt 0xc0221d7c <ehci_hub_control+2676>
0xc0221640 <ehci_hub_control+824>: sub r7, r3, #1 ; 0x1 0xc0221644 <ehci_hub_control+828>: add r3, r11, #16 ; 0x10 0xc0221648 <ehci_hub_control+832>: add r3, r8, r3, lsl #2 0xc022164c <ehci_hub_control+836>: uxth r7, r7 0xc0221650 <ehci_hub_control+840>: ldrb r5, [r3, #5] 0xc0221654 <ehci_hub_control+844>: ldrb r2, [r3, #4] 0xc0221658 <ehci_hub_control+848>: orr r5, r2, r5, lsl #8 0xc022165c <ehci_hub_control+852>: ldrb r2, [r3, #6] 0xc0221660 <ehci_hub_control+856>: ldrb r3, [r3, #7] 0xc0221664 <ehci_hub_control+860>: orr r5, r5, r2, lsl #16 0xc0221668 <ehci_hub_control+864>: orr r5, r5, r3, lsl #24 0xc022166c <ehci_hub_control+868>: dmb sy

As you can see, the old compiler accesses the ehci status register in a single access, the new compiler dances around and makes multiple accesses, which in the end get very incorrect data (I think that this register is like many which clear bits on reads).

Any ideas where I can go with this? I'm really trying to keep up with Poky/Yocto, but this move to GCC 4.6.0 has broken my ARM targets :-( I do have PowerPC targets as well - they seem fine (from limited testing)
with the new compiler.

Note: if you want to see the whole function disassembly, look at
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.5.2
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.6.0

Thanks

Quoted 9 lines
Quoted 4 lines

OK. thanks for sending the sample code.
Can you add -fstrict-volatile-bitfields option to CFLAGS while compiling this file or even the kernel for test sake ? and see if the problem goes away ?

Sorry, this doesn't seem to make any difference in the generated code. Hardly surprising as the register in question is a simple u32, not a bitfield.

I reduced the testcase a bit further and it seems the bug is similar to http://gcc.gnu.org/bugzilla/PR45819

but a bit different that in this case -fstrict-volatile-bitfields does not help

a fix for this particular problem is

include/linux/usb/ehci_def.h +1 -1
index 6563802..b8c1833 100644
@@ -194,7 +194,7 @@ struct ehci_dbg_port {
 u32 data47;
 u32 address;
 #define DBGP_EPADDR(dev, ep) (((dev)<<8)|(ep))
-} __attribute__ ((packed));
+} __attribute__ ((packed,aligned(__alignof__(int))));
 
 #ifdef CONFIG_EARLY_PRINTK_DBGP
 #include <linux/init.h>
Quoted 7 lines
Quoted 4 lines

since this particular structure all elements are integers but there might be more packed structures which have say char or short ints and are not necessarily ordered in the descending order of alignment

Try it out and see if it helps

Yes, this does fix the problem :-) The EHCI USB port on the OMAP is working again.

Quoted 5 lines

Note: it also works properly if the __attribute__ ((packed)) is simply removed. Looking at the layout of this structure, I don't see what that attribute is trying to accomplish - every item within the structure is u32[], so IMO there is nothing to pack. Perhaps this observation can help the GCC folks figure out why things are getting confused.

I tried each of these and neither worked. I confirmed the kernel was indeed different after each build and that gcc 4.6.0 was being used (by adding CC --version to the kernel do_compile and checking it in the log).

Gary, are you certain that you were building with 4.6.0 ? I've found it to be a bit tricky to switch back and forth between compilers.

Signature · 3 lines
Darren Hart
Intel Open Source Technology Center
Yocto Project - Linux Kernel
#12 lore raw
Quoted 139 lines · Darren Hart
Quoted 130 lines · Gary Thomas
Quoted 121 lines · Gary Thomas
Quoted 120 lines · Khem Raj
Quoted 88 lines · Gary Thomas
Quoted 87 lines · Khem Raj
Quoted 85 lines · Saul Wold
Quoted 10 lines · Gary Thomas

Now that Poky is using GCC 4.6.0, has anyone actually checked the operation on the BeagleBoard? I suspect that you'll find that the EHCI USB does not work. I can't really check here as I don't trust the EHCI on my BeagleBoard rev C3 (not xM).

That said, I've tried this new compiler (previously reported) on my own OMAP/3530 board which uses the same 2.6.37 kernel (just not the linux-yocto version). When I build my kernel with GCC 4.5.2, it works perfectly, but fails with GCC 4.6.0 (I trust the hardware).

Quoted 6 lines

I've isolated it down to at least the function 'ehci_hub_control' (but I suspect the problem is more fundamental). Comparing the code generated by the two compilers with the same source tree, this function is dramatically different. I can see why it's failing, I just don't know why the compiler is doing what it's doing.

Gary,

I am not sure if Khem Raj is on this list, so I am forwarding it to him, he might have some insight.

Sau!

Quoted 59 lines

The lines at drivers/usb/host/ehci-hub.c:841
case GetPortStatus:
if (!wIndex || wIndex> ports)
goto error;
wIndex--;
status = 0;
temp = ehci_readl(ehci, status_reg);

are being compiled very differently.

With GCC 4.5.2:
0xc0229810<ehci_hub_control+672>: cmp r3, #0 ; 0x0 0xc0229814<ehci_hub_control+676>: beq 0xc0229df8 <ehci_hub_control+2184> 0xc0229818<ehci_hub_control+680>: cmp r3, r0 0xc022981c<ehci_hub_control+684>: bgt 0xc0229df8 <ehci_hub_control+2184> 0xc0229820<ehci_hub_control+688>: sub r8, r3, #1 ; 0x1 0xc0229824<ehci_hub_control+692>: ldr r5, [r7, #4] 0xc0229828<ehci_hub_control+696>: uxth r8, r8 0xc022982c<ehci_hub_control+700>: dmb sy

With GCC 4.6.0:
0xc0221630<ehci_hub_control+808>: cmp r3, #0 ; 0x0 0xc0221634<ehci_hub_control+812>: beq 0xc0221d7c <ehci_hub_control+2676> 0xc0221638<ehci_hub_control+816>: cmp r3, r0 0xc022163c<ehci_hub_control+820>: bgt 0xc0221d7c <ehci_hub_control+2676> 0xc0221640<ehci_hub_control+824>: sub r7, r3, #1 ; 0x1 0xc0221644<ehci_hub_control+828>: add r3, r11, #16 ; 0x10 0xc0221648<ehci_hub_control+832>: add r3, r8, r3, lsl #2 0xc022164c<ehci_hub_control+836>: uxth r7, r7 0xc0221650<ehci_hub_control+840>: ldrb r5, [r3, #5] 0xc0221654<ehci_hub_control+844>: ldrb r2, [r3, #4] 0xc0221658<ehci_hub_control+848>: orr r5, r2, r5, lsl #8 0xc022165c<ehci_hub_control+852>: ldrb r2, [r3, #6] 0xc0221660<ehci_hub_control+856>: ldrb r3, [r3, #7] 0xc0221664<ehci_hub_control+860>: orr r5, r5, r2, lsl #16 0xc0221668<ehci_hub_control+864>: orr r5, r5, r3, lsl #24 0xc022166c<ehci_hub_control+868>: dmb sy

As you can see, the old compiler accesses the ehci status register in a single access, the new compiler dances around and makes multiple accesses, which in the end get very incorrect data (I think that this register is like many which clear bits on reads).

Any ideas where I can go with this? I'm really trying to keep up with Poky/Yocto, but this move to GCC 4.6.0 has broken my ARM targets :-( I do have PowerPC targets as well - they seem fine (from limited testing)
with the new compiler.

Note: if you want to see the whole function disassembly, look at
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.5.2
http://www.mlbassoc.com/poky/ehci_hub_control-disassembly-gcc-4.6.0

Thanks

Quoted 9 lines
Quoted 4 lines

OK. thanks for sending the sample code.
Can you add -fstrict-volatile-bitfields option to CFLAGS while compiling this file or even the kernel for test sake ? and see if the problem goes away ?

Sorry, this doesn't seem to make any difference in the generated code. Hardly surprising as the register in question is a simple u32, not a bitfield.

I reduced the testcase a bit further and it seems the bug is similar to http://gcc.gnu.org/bugzilla/PR45819

but a bit different that in this case -fstrict-volatile-bitfields does not help

a fix for this particular problem is

include/linux/usb/ehci_def.h +1 -1
index 6563802..b8c1833 100644
@@ -194,7 +194,7 @@ struct ehci_dbg_port {
 u32 data47;
 u32 address;
 #define DBGP_EPADDR(dev, ep) (((dev)<<8)|(ep))
-} __attribute__ ((packed));
+} __attribute__ ((packed,aligned(__alignof__(int))));
 
 #ifdef CONFIG_EARLY_PRINTK_DBGP
 #include<linux/init.h>
Quoted 7 lines
Quoted 4 lines

since this particular structure all elements are integers but there might be more packed structures which have say char or short ints and are not necessarily ordered in the descending order of alignment

Try it out and see if it helps

Yes, this does fix the problem :-) The EHCI USB port on the OMAP is working again.

Quoted 5 lines

Note: it also works properly if the __attribute__ ((packed)) is simply removed. Looking at the layout of this structure, I don't see what that attribute is trying to accomplish - every item within the structure is u32[], so IMO there is nothing to pack. Perhaps this observation can help the GCC folks figure out why things are getting confused.

Quoted 6 lines

I tried each of these and neither worked. I confirmed the kernel was indeed different after each build and that gcc 4.6.0 was being used (by adding CC --version to the kernel do_compile and checking it in the log).

Gary, are you certain that you were building with 4.6.0 ? I've found it to be a bit tricky to switch back and forth between compilers.

Absolutely - this is a clean, from scratch, build 100% GCC 4.6.0

Signature · 4 lines
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------
#13 lore raw
In reply to Darren Hart
Quoted 14 lines · Darren Hart
Quoted 10 lines · Gary Thomas
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines

The lines at drivers/usb/host/ehci-hub.c:841 case GetPortStatus:
if (!wIndex || wIndex > ports)
goto error;
wIndex--;
status = 0;
temp = ehci_readl(ehci, status_reg);

are being compiled very differently.

Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines
Quoted 11 lines

With GCC 4.5.2:
0xc0229810 <ehci_hub_control+672>: cmp r3, #0 ; 0x0 0xc0229814 <ehci_hub_control+676>: beq 0xc0229df8 <ehci_hub_control+2184>
0xc0229818 <ehci_hub_control+680>: cmp r3, r0 0xc022981c <ehci_hub_control+684>: bgt 0xc0229df8 <ehci_hub_control+2184>
0xc0229820 <ehci_hub_control+688>: sub r8, r3, #1 ; 0x1 0xc0229824 <ehci_hub_control+692>: ldr r5, [r7, #4] 0xc0229828 <ehci_hub_control+696>: uxth r8, r8 0xc022982c <ehci_hub_control+700>: dmb sy

Hi Gary, what exactly are you using to generate this output? objdump -d gives me something similar, but not exactly the same. With 4.6.0 with Khem's suggested patch:

include/linux/usb/ehci_def.h struct ehci_dbg_port { · line 200

Quoted 13 lines
include/linux/usb/ehci_def.h +1 -1
index 6563802..b8c1833 100644
@@ -194,7 +194,7 @@ struct ehci_dbg_port {
         u32     data47;
         u32     address;
  #define DBGP_EPADDR(dev, ep)   (((dev)<<8)|(ep))
-} __attribute__ ((packed));
+} __attribute__ ((packed,aligned(__alignof__(int))));
 
  #ifdef CONFIG_EARLY_PRINTK_DBGP
  #include <linux/init.h>

I see the following. Granted, this is a sligthly different kernel (includes linux-omap and the patches from the meta-ti linux-omap recipe):

$ /build/poky/master/beagleboard/tmp/work/armv7a-poky-linux-gnueabi/binutils-cross-2.21-r0/binutils-2.21/build.x86_64-linux.arm-poky-linux-gnueabi/binutils/objdump -S -d drivers/usb/host/ehci-hcd.o

<snip>

        case GetPortStatus:
                if (!wIndex || wIndex > ports)
    1420:       e3560000        cmp     r6, #0
    1424:       0a0001af        beq     1ae8 <ehci_hub_control+0xa40>
    1428:       e1560003        cmp     r6, r3
    142c:       ca0001ad        bgt     1ae8 <ehci_hub_control+0xa40>
    1430:       e2893010        add     r3, r9, #16
                        goto error;
                wIndex--;
    1434:       e2466001        sub     r6, r6, #1
    1438:       e0873103        add     r3, r7, r3, lsl #2
    143c:       e6ff6076        uxth    r6, r6
    1440:       e5d3b005        ldrb    fp, [r3, #5]
    1444:       e5d32004        ldrb    r2, [r3, #4]
    1448:       e182b40b        orr     fp, r2, fp, lsl #8
    144c:       e5d32006        ldrb    r2, [r3, #6]
    1450:       e5d33007        ldrb    r3, [r3, #7]
    1454:       e18bb802        orr     fp, fp, r2, lsl #16
    1458:       e18bbc03        orr     fp, fp, r3, lsl #24

#ifdef CONFIG_USB_EHCI_BIG_ENDIAN_MMIO
return ehci_big_endian_mmio(ehci) ?
readl_be(regs) :
readl(regs);
#else
return readl(regs);
145c: f57ff05f dmb sy

</snip>

This seems to demonstrate the same pile of extra ldrb and orr instructions that you first saw with 4.6.0. Some quick googling suggests Khem's idea of misaligned loads seems sound, but for whatever reason, the fix didn't work for me.

Khem, any other thoughts on something to try?

Note: CONFIG_USB_EHCI_BIG_ENDIAN_MMIO does not appear to be set in my
config.

Signature · 3 lines
Darren Hart
Intel Open Source Technology Center
Yocto Project - Linux Kernel
#14 lore raw
Quoted 9 lines · Darren Hart

This seems to demonstrate the same pile of extra ldrb and orr instructions that you first saw with 4.6.0. Some quick googling suggests Khem's idea of misaligned loads seems sound, but for whatever reason, the fix didn't work for me.

Khem, any other thoughts on something to try?

Note: CONFIG_USB_EHCI_BIG_ENDIAN_MMIO does not appear to be set in my
config.

If you can narrow the down the problem a bit may be I will get some direction otherwise I don't know what problem it might be. You config must be different to Gary's and there could be same or different problem
in other part of code.

Give me more information on the error and ideally like what Gary did makes it easy