<div dir="ltr"><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Fri, Jan 18, 2019 at 8:22 PM Bas Nieuwenhuizen <<a href="mailto:bas@basnieuwenhuizen.nl">bas@basnieuwenhuizen.nl</a>> wrote: </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">On Sat, Jan 19, 2019 at 2:10 AM Marek Olšák <<a href="mailto:maraeo@gmail.com" target="_blank">maraeo@gmail.com</a>> wrote: > > On Fri, Jan 18, 2019 at 6:08 PM Bas Nieuwenhuizen <<a href="mailto:bas@basnieuwenhuizen.nl" target="_blank">bas@basnieuwenhuizen.nl</a>> wrote: >> >> On Fri, Jan 18, 2019 at 5:44 PM Marek Olšák <<a href="mailto:maraeo@gmail.com" target="_blank">maraeo@gmail.com</a>> wrote: >> > >> > From: Marek Olšák <<a href="mailto:marek.olsak@amd.com" target="_blank">marek.olsak@amd.com</a>> >> > >> > --- >> > 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)? </blockquote><div> </div><div>True. I'll probably drop the WRITE_DATA optimization, because it complicates the code with an uncertain benefit.</div><div> </div><div>Marek </div></div></div>