[Bug bootstrap/93074] [10 regression] build FAIL with --enable-offload-targets=nvptx-none

2019-12-28 Thread jakub at gcc dot gnu.org 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93074

Jakub Jelinek  changed:

   What|Removed |Added

 Status|UNCONFIRMED |RESOLVED
 Resolution|--- |FIXED
   Assignee|unassigned at gcc dot gnu.org  |jakub at gcc dot gnu.org
   Target Milestone|--- |10.0

--- Comment #6 from Jakub Jelinek  ---
Should be fixed now.

[Bug bootstrap/93074] [10 regression] build FAIL with --enable-offload-targets=nvptx-none

2019-12-28 Thread jakub at gcc dot gnu.org 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93074

--- Comment #5 from Jakub Jelinek  ---
Author: jakub
Date: Sat Dec 28 09:26:03 2019
New Revision: 279747

URL: https://gcc.gnu.org/viewcvs?rev=279747=gcc=rev
Log:
PR bootstrap/93074
* plugin/cuda/cuda.h (cuDeviceGetName, cuDriverGetVersion): Declare.
(cuDeviceTotalMem, cuMemGetInfo): Likewise.  Define to *_v2.

Modified:
trunk/libgomp/ChangeLog
trunk/libgomp/plugin/cuda/cuda.h

[Bug bootstrap/93074] [10 regression] build FAIL with --enable-offload-targets=nvptx-none

2019-12-27 Thread tschwinge at gcc dot gnu.org 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93074

Thomas Schwinge  changed:

   What|Removed |Added

 CC||frederik at gcc dot gnu.org,
   ||tschwinge at gcc dot gnu.org

--- Comment #4 from Thomas Schwinge  ---
For Frederik (CCed): this is the GCC configuration where libgomp doesn't
compile/link against a CUDA Driver installation, but instead uses its own stub
header file ('libgomp/plugin/cuda/cuda.h') and then 'dlopen's 'libcuda.so', and
the former needs to be updated per the new interfaces you're using in r279710
"Add OpenACC 2.6 `acc_get_property' support".

[Bug bootstrap/93074] [10 regression] build FAIL with --enable-offload-targets=nvptx-none

2019-12-27 Thread jakub at gcc dot gnu.org 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93074

Jakub Jelinek  changed:

   What|Removed |Added

 CC||jakub at gcc dot gnu.org

--- Comment #3 from Jakub Jelinek  ---
I think we need:
--- libgomp/plugin/cuda/cuda.h.jj   2019-05-14 21:37:34.997366009 +0200
+++ libgomp/plugin/cuda/cuda.h  2019-12-27 11:13:38.078883739 +0100
@@ -126,8 +126,11 @@ CUresult cuCtxPopCurrent (CUcontext *);
 CUresult cuCtxPushCurrent (CUcontext);
 CUresult cuCtxSynchronize (void);
 CUresult cuDeviceGet (CUdevice *, int);
+#define cuDeviceTotalMem cuDeviceTotalMem_v2
+CUresult cuDeviceTotalMem (size_t *, CUdevice);
 CUresult cuDeviceGetAttribute (int *, CUdevice_attribute, CUdevice);
 CUresult cuDeviceGetCount (int *);
+CUresult cuDeviceGetName (char *, int, CUdevice);
 CUresult cuEventCreate (CUevent *, unsigned);
 #define cuEventDestroy cuEventDestroy_v2
 CUresult cuEventDestroy (CUevent);
@@ -138,6 +141,7 @@ CUresult cuEventSynchronize (CUevent);
 CUresult cuFuncGetAttribute (int *, CUfunction_attribute, CUfunction);
 CUresult cuGetErrorString (CUresult, const char **);
 CUresult cuInit (unsigned);
+CUresult cuDriverGetVersion (int *);
 CUresult cuLaunchKernel (CUfunction, unsigned, unsigned, unsigned, unsigned,
 unsigned, unsigned, unsigned, CUstream, void **, void
**);
 #define cuLinkAddData cuLinkAddData_v2
@@ -147,6 +151,8 @@ CUresult cuLinkComplete (CUlinkState, vo
 #define cuLinkCreate cuLinkCreate_v2
 CUresult cuLinkCreate (unsigned, CUjit_option *, void **, CUlinkState *);
 CUresult cuLinkDestroy (CUlinkState);
+#define cuMemGetInfo cuMemGetInfo_v2
+CUresult cuMemGetInfo (size_t *, size_t *);
 #define cuMemAlloc cuMemAlloc_v2
 CUresult cuMemAlloc (CUdeviceptr *, size_t);
 #define cuMemAllocHost cuMemAllocHost_v2
but will test and verify it only in the evening.

[Bug bootstrap/93074] [10 regression] build FAIL with --enable-offload-targets=nvptx-none

2019-12-26 Thread dimhen at gmail dot com 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93074

--- Comment #2 from Dmitry G. Dyachenko  ---
(In reply to Andrew Pinski from comment #1)
> According to
> https://docs.nvidia.com/cuda/cuda-driver-api/group__CUDA__DEVICE.html
> 
> cuDeviceGetName exists.
> Maybe F31 has an older version of Cuda installed.

I have no CUDA installed.
Can I check smth else?

r279710 FAIL
r279709 PASS

[Bug bootstrap/93074] [10 regression] build FAIL with --enable-offload-targets=nvptx-none

2019-12-25 Thread pinskia at gcc dot gnu.org 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93074

--- Comment #1 from Andrew Pinski  ---
According to
https://docs.nvidia.com/cuda/cuda-driver-api/group__CUDA__DEVICE.html

cuDeviceGetName exists.
Maybe F31 has an older version of Cuda installed.