On Fri, Jan 23, 2015 at 03:33:32PM +0100, Michal Simek wrote:
On 01/23/2015 03:12 PM, Tom Rini wrote:
On Fri, Jan 23, 2015 at 02:39:44PM +0100, Michal Simek wrote:
On 01/23/2015 12:59 PM, Tom Rini wrote:
On Fri, Jan 23, 2015 at 10:02:04AM +0100, Michal Simek wrote:
On 01/23/2015 02:05 AM, Tom Rini wrote:
On Wed, Jan 21, 2015 at 10:38:47AM +0100, Michal Simek wrote:
> Hi Tom, > > please pull these patches to your tree. I have added there 2 patches > for serial. Zynq is only one platform which is using this driver. > > Thanks, > Michal > > The following changes since commit 92fa7f53f1f3f03296f8ffb14bdf1baefab83368: > > Prepare v2015.01 (2015-01-12 09:39:08 -0500) > > are available in the git repository at: > > git://www.denx.de/git/u-boot-microblaze.git zynq > > for you to fetch changes up to 025216f78e0dacd491908d0d76a3642630cd84dc: > > serial: Extend structure comments with register offset (2015-01-21 10:36:36 +0100) >
I can't find a toolchain that doesn't give me something like: arm: + zynq_zc70x +(zynq_zc70x) arch/arm/cpu/armv7/zynq/lowlevel_init.S: Assembler messages: +(zynq_zc70x) arch/arm/cpu/armv7/zynq/lowlevel_init.S:19: Error: selected processor doe s not support ARM mode `fmrx r1,FPEXC' +(zynq_zc70x) arch/arm/cpu/armv7/zynq/lowlevel_init.S:21: Error: selected processor does not support ARM mode `fmxr FPEXC,r1' +(zynq_zc70x) make[3]: *** [arch/arm/cpu/armv7/zynq/lowlevel_init.o] Error 1 +(zynq_zc70x) make[2]: *** [arch/arm/cpu/armv7/zynq] Error 2 +(zynq_zc70x) make[1]: *** [arch/arm/cpu/armv7] Error 2 +(zynq_zc70x) make: *** [sub-make] Error 2
ok. I see. We have neon instructions enabled by default in xilinx toolchain. I have sent the patch which create and add PLATFORM_RELFLAGS += -mfpu=neon to config.mk.
This should solve the problem with compilation. No problem to add this patch as the first in this serial not to break bisectability of the tree.
And we have a _really_ good reason for using FPU instructions, yes?
The description for doing that is in the patch but to be honest the biggest problem is that xilinx toolchain has FPU instructions enabled by default. Based on that fpu instructions can be used. For standard xilinx flow FSBL/u-boot - fsbl does that setting. For debug flow TCL + u-boot we are reaching this issue.
Maybe the second option can be to disable FPU instructions for u-boot compilation but then the problem is moved to next software which is designed to have FPU instructions enabled. That's why I think it is just better to enable it in u-boot.
Siva: Feel free to correct me if my understanding is not correct.
And for the record and clarity, we're still not actually using the FPU in U-Boot nor intend to, just enabling normal workflows on these platforms, right?
Yes, correct.
OK, good, waiting for the new PR then, thanks!