On Fri, Jan 18, 2019 at 8:22 PM Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
wrote:
> On Sat, Jan 19, 2019 at 2:10 AM Marek Olšák <mar...@gmail.com> wrote:
> >
> > On Fri, Jan 18, 2019 at 6:08 PM Bas Nieuwenhuizen <
> b...@basnieuwenhuizen.nl> wrote:
> >>
> >> On Fri, Jan 18, 2019 at 5:44 PM Marek Olšák <mar...@gmail.com> wrote:
> >> >
> >> > From: Marek Olšák <marek.ol...@amd.com>
> >> >
> >> > ---
> >> > src/gallium/auxiliary/Makefile.sources | 1 +
> >> > src/gallium/auxiliary/meson.build | 1 +
> >> > src/gallium/auxiliary/util/u_cpu_suballoc.h | 90
> +++++++++++++++++++++
> >> > 3 files changed, 92 insertions(+)
> >> > create mode 100644 src/gallium/auxiliary/util/u_cpu_suballoc.h
> >> >
> >> > diff --git a/src/gallium/auxiliary/Makefile.sources
> b/src/gallium/auxiliary/Makefile.sources
> >> > index 50e88088ff8..b26415858f6 100644
> >> > --- a/src/gallium/auxiliary/Makefile.sources
> >> > +++ b/src/gallium/auxiliary/Makefile.sources
> >> > @@ -211,20 +211,21 @@ C_SOURCES := \
> >> > util/u_bitmask.c \
> >> > util/u_bitmask.h \
> >> > util/u_blend.h \
> >> > util/u_blit.c \
> >> > util/u_blit.h \
> >> > util/u_blitter.c \
> >> > util/u_blitter.h \
> >> > util/u_box.h \
> >> > util/u_cache.c \
> >> > util/u_cache.h \
> >> > + util/u_cpu_suballoc.h \
> >> > util/u_debug_gallium.h \
> >> > util/u_debug_gallium.c \
> >> > util/u_debug_describe.c \
> >> > util/u_debug_describe.h \
> >> > util/u_debug_flush.c \
> >> > util/u_debug_flush.h \
> >> > util/u_debug_image.c \
> >> > util/u_debug_image.h \
> >> > util/u_debug_memory.c \
> >> > util/u_debug_refcnt.c \
> >> > diff --git a/src/gallium/auxiliary/meson.build
> b/src/gallium/auxiliary/meson.build
> >> > index 57f7e69050f..7e1e4732421 100644
> >> > --- a/src/gallium/auxiliary/meson.build
> >> > +++ b/src/gallium/auxiliary/meson.build
> >> > @@ -231,20 +231,21 @@ files_libgallium = files(
> >> > 'util/u_bitmask.c',
> >> > 'util/u_bitmask.h',
> >> > 'util/u_blend.h',
> >> > 'util/u_blit.c',
> >> > 'util/u_blit.h',
> >> > 'util/u_blitter.c',
> >> > 'util/u_blitter.h',
> >> > 'util/u_box.h',
> >> > 'util/u_cache.c',
> >> > 'util/u_cache.h',
> >> > + 'util/u_cpu_suballoc.h',
> >> > 'util/u_debug_gallium.h',
> >> > 'util/u_debug_gallium.c',
> >> > 'util/u_debug_describe.c',
> >> > 'util/u_debug_describe.h',
> >> > 'util/u_debug_flush.c',
> >> > 'util/u_debug_flush.h',
> >> > 'util/u_debug_image.c',
> >> > 'util/u_debug_image.h',
> >> > 'util/u_debug_memory.c',
> >> > 'util/u_debug_refcnt.c',
> >> > diff --git a/src/gallium/auxiliary/util/u_cpu_suballoc.h
> b/src/gallium/auxiliary/util/u_cpu_suballoc.h
> >> > new file mode 100644
> >> > index 00000000000..2373c1f7c70
> >> > --- /dev/null
> >> > +++ b/src/gallium/auxiliary/util/u_cpu_suballoc.h
> >> > @@ -0,0 +1,90 @@
> >> >
> +/**************************************************************************
> >> > + *
> >> > + * Copyright 2019 Advanced Micro Devices, Inc.
> >> > + * All Rights Reserved.
> >> > + *
> >> > + * Permission is hereby granted, free of charge, to any person
> obtaining a
> >> > + * copy of this software and associated documentation files (the
> >> > + * "Software"), to deal in the Software without restriction,
> including
> >> > + * without limitation the rights to use, copy, modify, merge,
> publish,
> >> > + * distribute, sub license, and/or sell copies of the Software, and
> to
> >> > + * permit persons to whom the Software is furnished to do so,
> subject to
> >> > + * the following conditions:
> >> > + *
> >> > + * The above copyright notice and this permission notice (including
> the
> >> > + * next paragraph) shall be included in all copies or substantial
> portions
> >> > + * of the Software.
> >> > + *
> >> > + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
> EXPRESS
> >> > + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
> >> > + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
> NON-INFRINGEMENT.
> >> > + * IN NO EVENT SHALL AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR
> >> > + * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF
> CONTRACT,
> >> > + * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
> >> > + * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
> >> > + *
> >> > +
> **************************************************************************/
> >> > +
> >> > +/* A simple utility for suballocating out of malloc_aligned. */
> >> > +
> >> > +#ifndef U_CPU_SUBALLOC_H
> >> > +#define U_CPU_SUBALLOC_H
> >> > +
> >> > +#include <stdint.h>
> >> > +#include "util/os_memory.h"
> >> > +
> >> > +struct u_cpu_suballoc {
> >> > + unsigned default_size; /* Default size of the buffer, in bytes.
> */
> >> > + unsigned current_size; /* Current size of the buffer, in bytes.
> */
> >> > + unsigned alignment; /* malloc alignment. */
> >> > + unsigned offset; /* Offset pointing to the first unused
> byte. */
> >> > + uint8_t *buffer; /* Pointer to the CPU buffer. */
> >> > +};
> >> > +
> >> > +
> >> > +static inline void
> >> > +u_cpu_suballoc_init(struct u_cpu_suballoc *alloc, unsigned
> default_size,
> >> > + unsigned alignment)
> >> > +{
> >> > + memset(alloc, 0, sizeof(*alloc));
> >> > + alloc->default_size = default_size;
> >> > + alloc->alignment = alignment;
> >> > +}
> >> > +
> >> > +
> >> > +static inline void
> >> > +u_cpu_suballoc_deinit(struct u_cpu_suballoc *alloc)
> >> > +{
> >> > + os_free_aligned(alloc->buffer);
> >> > + alloc->buffer = NULL;
> >> > +}
> >> > +
> >> > +
> >> > +static inline void *
> >> > +u_cpu_suballoc(struct u_cpu_suballoc *alloc, unsigned size, unsigned
> alignment)
> >> > +{
> >> > + unsigned offset = align(alloc->offset, alignment);
> >> > +
> >> > + /* Make sure we have enough space in the buffer for the
> sub-allocation. */
> >> > + if (unlikely(!alloc->buffer || offset + size >
> alloc->current_size)) {
> >> > + os_free_aligned(alloc->buffer);
> >>
> >> So, if we can free the memory during a suballocation, how do we make
> >> sure that the previous allocations are valid? If they don't have to
> >> stay valid, why use a linear allocator at all?
> >
> >
> > The assumption is that previously allocated memory won't be accessed
> after the next call to u_cpu_suballoc. Yeah it's pretty fragile.
>
> What is the point then of having this as suballocator? If we just want
> to cache a memory allocation, why not keep offset at 0 (or rather
> remove it completely)?
>
True. I'll probably drop the WRITE_DATA optimization, because it
complicates the code with an uncertain benefit.
Marek
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev
