use #include instead of soft links from .cu to .c files - as links are causing grief with mingw

Reported-by: Pierre Jolivet <pierre.jolivet@enseeiht.fr>

Merge remote-tracking branch 'origin/maint'

PCView error with MUMPS, GAMG, and LMVM

Merge remote-tracking branch 'origin/stefanozampini/fix-matshell-axpy'

Fix MATSHELL MatAXPY

See merge request petsc/petsc!2492

Changed XXXPinToCPU() to XXXBindToCPU() to prevent confusion.

The reason for this change is that we already use the terminology
"pinned" to refer to memory that is non-pageable, in the context of
Pe

Changed XXXPinToCPU() to XXXBindToCPU() to prevent confusion.

The reason for this change is that we already use the terminology
"pinned" to refer to memory that is non-pageable, in the context of
PetscSF as well as allocating host memory when GPUs are being employed.

show more ...

MatCopy_SeqAIJ: error diagnostics

General PETSc fixes

* make generatefortranstubs.py handle .cxx
* bump HPDDM version + minor hypre cleaning

Merge remote-tracking branch 'origin/maint'

Activate PARDISO for complex symmetric matrices

Merge remote-tracking branch 'origin/maint'

Double CHKERRQ

Merge remote-tracking branch 'origin/maint'

Add logic setting offloadmask for sequential matrices in the MPIAIJ code that mimicks logic in MatSetValues_SeqAIJ().

Merge remote-tracking branch 'origin/maint'

bugfix for MatMumpsGetInverse_MUMPS()
Reported-by: Jan Grießer <griesser.jan@googlemail.com>

Merge remote-tracking branch 'origin/maint'

MatDuplicate_MPIAIJ: fix for the case of unassembled matrices

Merge remote-tracking branch 'origin/maint'

Fix nonscalable code in PCTELESCOPE

Propagate flag in MatPermute_SeqAIJ

MATSUPERLUDIST: fix MPIAIJ with commsize 1 case

MatMPIAIJGetLocalMat: fix reuse case and clarify man page

This "optimization" was introduced in 70a9ba44dd1.
There's no need to create a new matrix with MatCreateSeqAIJWithArrays since it will share

MatMPIAIJGetLocalMat: fix reuse case and clarify man page

This "optimization" was introduced in 70a9ba44dd1.
There's no need to create a new matrix with MatCreateSeqAIJWithArrays since it will share the underlying data structure
Better off to just take a reference on the local matrix and return it.
Extended the MAT_REUSE_CASE (MatCopy is a dummy operation if the two matrices are the same one): at least in this case we get an error if we are reusing the wrong matrix (sparsity pattern changed)

show more ...

MATCHOLMOD: support for MatMatSolve and MatGetInfo

Fix bug with complex matrices: cholmod requires hermitian matrices (upper triangular)
Add missing prefix for options: all the other factorization p

MATCHOLMOD: support for MatMatSolve and MatGetInfo

Fix bug with complex matrices: cholmod requires hermitian matrices (upper triangular)
Add missing prefix for options: all the other factorization packages search for options prefixed by the matrix prefix

KSP ex62: add umfpack test and disable few cases for complex

show more ...

MATSBAIJ: Fix Hermitian MatMult for Seq and MPI code paths

we can reuse the same MatMult operation with no impact for non-complex builds
Remove improper check for the hermitian complex case from Mat

MATSBAIJ: Fix Hermitian MatMult for Seq and MPI code paths

we can reuse the same MatMult operation with no impact for non-complex builds
Remove improper check for the hermitian complex case from MatGetFactor and move it where it actually belongs
Fix conversion routines from SBAIJ when the hermitian flag is set

show more ...

minor