Commit Graph

2518 Commits

Author SHA1 Message Date
Alastair Donaldson
94d6002dc5
spirv-fuzz: Pass on validator options during shrinking (#3317)
Fixes an issue where the shrinker's validator options were not
actually being passed to the validator.
2020-04-27 14:25:46 +01:00
Alastair Donaldson
88faf63ad3
spirv-fuzz: Clamp statically out-of-bounds accesses in code donation (#3315)
It has been resolved that statically out-of-bounds accesses are not
invalid in SPIR-V (they lead to undefind behaviour at runtime but
should not cause a module to be rejected during validation).  This
change tolerates such accesses in donated code, clamping them in-bound
as part of making a function live-safe.
2020-04-27 14:24:54 +01:00
Alastair Donaldson
b74199a22d
spirv-fuzz: Fix memory management in the fact manager (#3313)
Fixes a bug where, while recursively adding id equation facts, a
reference to a set of id equations could be used after it had been
freed (due to equivalence classes of equations being merged).
2020-04-27 14:24:11 +01:00
Alastair Donaldson
d158ffe540
spirv-fuzz: Do not replace the Sample argument in OpImageTexelPointer (#3311)
The Sample argument of OpImageTexelPointer is sometimes required to be
a zero constant.  It thus cannot be replaced with a synonym in
general.  This change avoids replacing this argument with a synonym.
2020-04-27 14:23:11 +01:00
Alastair Donaldson
5547553a0c
Allow various validation options to be passed to spirv-opt (#3314)
This change increases the set of validator options that can be passed
to spirv-opt, to match those options that spirv-reduce and spirv-fuzz
accept.  This is useful to still allow some validation, at the start
of and during optimisation, for SPIR-V modules that the strict
validator would reject.
2020-04-23 16:25:29 +01:00
CHAITANYA
30ffe62e25
typo fix: in README.md exectuable->executable (#3306) 2020-04-23 10:56:47 -04:00
Alastair Donaldson
67f4838659
spirv-fuzz: Make handling of synonym facts more efficient (#3301)
The fact manager maintains an equivalence relation on data descriptors
that tracks when one data descriptor could be used in place of
another.  An algorithm to compute the closure of such facts allows
deducing new synonym facts from existing facts.  E.g., for two 2D
vectors u and v it is known that u.x is synonymous with v.x and u.y is
synonymous with v.y, it can be deduced that u and v are synonymous.

The closure computation algorithm is very expensive if we get large
equivalence relations.

This change addresses this in three ways:

- The size of equivalence relations is reduced by limiting the extent
  to which the components of a composite are recursively noted as
  being equivalent, so that when we have large synonymous arrays we do
  not record all array elements as being pairwise equivalent.

- When computing the closure of facts, equivalence classes above a
  certain size are simply skipped (which can lead to missed facts)

- The closure computation is performed less frequently - it is invoked
  explicitly before fuzzer passes that will benefit from data synonym
  facts.  A new transformation is used to control its invocation, so
  that fuzzing and replaying do not get out of sync.

The change also tidies up the order in which some getters are declared
in FuzzerContext.
2020-04-20 19:02:49 +01:00
Steven Perron
61b7de3c39
Remove unreachable code. (#3304) 2020-04-15 14:41:52 -04:00
Alastair Donaldson
ed96301c6c
spirv-fuzz: Fix to outliner (#3302)
Adds an extra condition on when a region can be outlined to avoid the
case where a region ends with a loop head but such that the loop's
continue target is in the region.  (Outlining such a region would mean
that the loop merge is in the original function and the continue target
in the outlined function.)
2020-04-15 11:39:33 +01:00
Alastair Donaldson
c018fc6ae6
spirv-fuzz: Do not outline regions that produce pointer outputs (#3291)
The function outliner uses a struct to return ids that a region
generates and that are used outside that region.  If these ids have
pointer type this would result in a struct with pointer members, which
leads to illegal loading from non-logical pointers if logical
addressing is used.  This change bans that outlining possibility.
2020-04-14 20:22:30 +01:00
Alastair Donaldson
f460cca9dc
spirv-fuzz: Handle OpRuntimeArray when replacing ids with synonyms (#3292)
Provides support for runtime arrays in the code that traverses
composite types when checking applicability of transformations that
replace ids with synonyms.
2020-04-14 20:21:58 +01:00
Alastair Donaldson
2f180468a7
spirv-fuzz: Handle image storage class in donation (#3290)
Demotes the image storage class to Private during donation.  Also
fixes an issue where instructions that depended on non-donated global
values would not be handled properly.
2020-04-14 20:21:33 +01:00
Alastair Donaldson
f82d47003e
spirv-fuzz: Respect rules for OpSampledImage (#3287)
The SPIR-V data rules say that all uses of an OpSampledImage
instruction must be in the same block as the instruction, and highly
restrict those instructions that can consume the result id of an
OpSampledImage.

This adapts the transformations that split blocks and create synonyms
to avoid separating an OpSampledImage use from its definition, and to
avoid synonym-creation instructions such as OpCopyObject consuming an
OpSampledImage result id.
2020-04-14 20:17:42 +01:00
Alastair Donaldson
7ce2db1763
spirv-fuzz: Fix comment. (#3300) 2020-04-14 20:17:05 +01:00
Steven Perron
7d65bce0bb
Sampled images as read-only storage (#3295)
There are some cases where a variable that is declared as a sampled
image could be read only.  That is when the image type has sampled == 1.

Fixes #3288
2020-04-14 12:58:05 -04:00
alan-baker
2a2bdbd5d7
Remove implicit fallthrough (#3298)
Fixes #3296

* Make OpReturn its own case fully
2020-04-14 12:57:48 -04:00
Steven Perron
4956644894
Add tests for recently added command line option (#3297)
We have not added tests for the new command line options recently.  I've
updated the test and fixed on option that was incorrect.

Fixes #3247
2020-04-14 12:57:06 -04:00
David Neto
ca5751590e
If SPIRV-Headers is in our tree, include it as subproject (#3299)
This allows enclosing projects to use SPIRV-Headers_SOURCE_DIR
to set up in header includes.
2020-04-14 11:28:25 -04:00
Steven Perron
e70d25f6fa
Struct CFG analysus and single block loop (#3293)
Loop headers must be marked as in the continue if the loop header is
also the continue target.

Fixes #3264
2020-04-13 10:08:31 -04:00
Jaebaek Seo
000040e707
Preserve debug info in eliminate-dead-functions (#3251)
* Preserve debug info in eliminate-dead-functions

The elimination of dead functions makes OpFunction operand of
DebugFunction invalid. This commit replaces the operand with
DebugInfoNone.
2020-04-13 09:29:36 -04:00
Steven Perron
c531099eb3
Update acorn version (#3294) 2020-04-13 09:24:47 -04:00
Steven Perron
34be23373b
Handle more cases in dead member elim (#3289)
* Handle more cases in dead member elim

- Rewrite composite insert and extract operations on SpecConstnatOp.
- Leaves assert for Access chain instructions, which are only allowed
for kernels.
- Other operations do not require any extra code will no longer cause an
assert.

Fixes #3284.
Fixes #3282.
2020-04-09 15:44:20 -04:00
Malacath-92
d0490ef080
Fix pch macro to ignore clang-cl (#3283) 2020-04-09 14:46:08 -04:00
Alastair Donaldson
538512e8e8
spirv-fuzz: Improve the handling of equation facts (#3281)
The management of equation facts suffered from two problems:

(1) The processing of an equation fact required the data descriptors
    used in the equation to be in canonical form.  However, during
    fact processing it can be deduced that certain data descriptors
    are equivalent, causing their equivalence classes to be merged,
    and that could cause previously canonical data descriptors to no
    longer be canonical.

(2) Related to this, if id equations were known about a canonical data
    descriptor dd1, and other id equations known about a different
    canonical data descriptor dd2, the equation facts about these data
    descriptors were not being merged in the event that dd1 and dd2
    were deduced to be equivalent.

This changes solves (1) by not requiring equation facts to be in
canonical form while processing them, but instead always checking
whether (not necessary canonical) data descriptors are equivalent when
looking for corollaries of equation facts, rather than comparing them
using ==.

Problem (2) is solved by adding logic to merge sets of equations when
data descriptors are made equivalent.

In addition, the change also requires elements to be registered in an
equivalence relation before they can be made equivalent, rather than
being added (if not already present) at the point of being made
equivalent.
2020-04-07 17:38:27 +01:00
Alastair Donaldson
183e3242a3
spirv-fuzz: Handle more general SPIR-V in donation (#3280)
This change increases the extent to which arbitrary SPIR-V can be used
by the fuzzer pass that donates modules. It handles the case where
various ingredients (such as types, variables and particular
instructions) cannot be donated by omitting them, and then either
omitting their dependencies or replacing their dependencies with
alternative instructions.

The change pays particular attention to allowing code that manipulates
image types to be handled (by skipping anything image-specific).
2020-04-07 17:37:51 +01:00
Alastair Donaldson
4af38c49bf
spirv-fuzz: Improve support for compute shaders in donation (#3277)
(1) Runtime arrays are turned into fixed-size arrays, by turning
    OpTypeRuntimeArray into OpTypeArray and uses of OpArrayLength into
    uses of the constant used for the length of the fixed-size array.

(2) Atomic instructions are not donated, and uses of their results are
    replaced with uses of constants of the result type.
2020-04-06 16:08:14 +01:00
Alastair Donaldson
e95fbfb1f5
spirv-fuzz: Transformation to add OpConstantNull (#3273)
Adds a transformation for adding OpConstantNull to a module, for
appropriate data types.
2020-04-02 19:25:30 +01:00
Alastair Donaldson
5d491a7ed6
spirv-fuzz: Handle isomorphic types property in composite construction (#3262)
The fuzzer pass that constructs composites had an issue where it would
regard isomorphic but distinct structs (similarly arrays) as being
interchangeable when constructing composites.  This change fixes the
problem by relying less on the type manager.
2020-04-02 18:33:19 +01:00
Alastair Donaldson
bfd25ace08
spirv-fuzz: Limit adding of new variables to 'basic' types (#3257)
To avoid problems where global and local variables of opaque or
runtime-sized types are added to a module, this change introduces the
notion of a 'basic type' -- a type made up from floats, ints, bools,
or vectors, matrices, structs and fixed-size arrays of basic types.
Added variables have to be of basic type.
2020-04-02 17:35:18 +01:00
Alastair Donaldson
f28cdeff16
spirv-fuzz: Only replace regular ids with synonyms (#3255)
Fixes an issue where an id use in a non-regular context, e.g. as a
scope parameter to an atomic instruction, would be replaced with a
synonym.
2020-04-02 17:35:00 +01:00
Alastair Donaldson
8d4261bc44
spirv-fuzz: Introduce TransformationContext (#3272)
Some transformations (e.g. TransformationAddFunction) rely on running
the validator to decide whether the transformation is applicable.  A
recent change allowed spirv-fuzz to take validator options, to cater
for the case where a module should be considered valid under
particular conditions.  However, validation during the checking of
transformations had no access to these validator options.

This change introduced TransformationContext, which currently consists
of a fact manager and a set of validator options, but could in the
future have other fields corresponding to other objects that it is
useful to have access to when applying transformations.  Now, instead
of checking and applying transformations in the context of a
FactManager, a TransformationContext is used.  This gives access to
the fact manager as before, and also access to the validator options
when they are needed.
2020-04-02 15:54:46 +01:00
Alastair Donaldson
2fdea57d19
spirv-fuzz: Add validator options (#3254)
Allows several validator options to be passed to the fuzzer, to be
used when validation is invoked during fuzzing.
2020-04-02 14:37:59 +01:00
alan-baker
af01d57b5e
Update dominates to check for null nodes (#3271)
* Update dominates to check for null nodes

Fixes #3270
2020-04-02 08:19:54 -04:00
alan-baker
f20c0d7971
Set wrapped kill basic block's parent (#3269)
Fixes #3268

* Set the parent of the basic block for the wrapper function
* Add a test
2020-04-01 12:31:57 -04:00
Caio Marcelo de Oliveira Filho
c37c94929b
Validate Buffer and BufferBlock apply only to struct types (#3259) 2020-03-31 10:06:29 -04:00
Steven Perron
fd773eb50d Start SPIRV-Tools v2020.3 2020-03-26 02:04:07 -04:00
Steven Perron
fd8e130510 Finalize SPIRV-Tools v2020.2 2020-03-26 02:02:59 -04:00
Steven Perron
3ef8fe9a52 Update CHANGES 2020-03-26 02:01:43 -04:00
alan-baker
022da4d0e0
Fix identification of Vulkan images and buffers (#3253)
Fixes #3252

* Image and buffer queries did not account for optional level of
arrayness on the variable
  * new tests
2020-03-25 17:38:24 -04:00
alan-baker
1346dd5de1
Disallow phis of images, samplers and sampled images (#3246)
* Disallow phis of sampled images always
* Disallow phis of samplers and images for shaders
* Add tests
* Gate check to only occur post-legalization
2020-03-23 16:59:37 -04:00
Jaebaek Seo
1c8bda3721
Add data structure for DebugScope, DebugDeclare in spirv-opt (#3183)
When DebugScope is given in SPIR-V, each instruction following the
DebugScope is from the lexical scope pointed by the DebugScope in
the high level language. We add DebugScope struction to keep the
scope information in Instruction class. When ir_loader loads
DebugScope/DebugNoScope, it keeps the scope information in
|last_dbg_scope_| and lets following instructions have that scope
information.

In terms of DebugDeclare/DebugValue, if it is in a function body
but outside of a basic block, we keep it in |debug_insts_in_header_|
of Function class. If it is in a basic block, we keep it as a normal
instruction i.e., in a instruction list of BasicBlock.
2020-03-23 11:01:18 -04:00
Ehsan
e6f372c5c2
Whitelist SPV_KHR_ray_tracing (#3241) 2020-03-23 09:31:05 -05:00
Aaron Franke
7f341ffee6
Make file formatting comply with POSIX standards (#3242)
Add newlines to two files
2020-03-23 10:20:00 -04:00
David Neto
60104cd974
Add opt::Operand::AsCString and AsString (#3240)
It only works when the operand is a literal string.
2020-03-19 12:44:28 -04:00
JiaoluAMD
da52d0875c
Add RayQueryProvisionalKHR to opt types (#3239)
Add missing RayQueryProvisionalKHR types
2020-03-19 12:41:30 -04:00
Ehsan
18d3896a15
Whitelist SPV_EXT_demote_to_helper_invocation for opt passes (#3236) 2020-03-17 22:36:02 -05:00
Daniel Koch
5a97e3a391
Add support for KHR_ray_{query,tracing} extensions (#3235)
Update validator for SPV_KHR_ray_tracing.

* Added handling for new enum types
* Add SpvScopeShaderCallKHR as a valid scope
* update spirv-headers

Co-authored-by: alelenv <alele@nvidia.com>
Co-authored-by: Torosdagli <ntorosda@amd.com>
Co-authored-by: Tobias Hector <tobias.hector@amd.com>
Co-authored-by: Steven Perron <stevenperron@google.com>
2020-03-17 15:30:19 -04:00
Jamie Madill
25ede1ced6
Roll external/spirv-headers/ 30ef660ce..a17e17e36 (1 commit) (#3230)
30ef660ce2...a17e17e36d

$ git log 30ef660ce..a17e17e36 --date=short --no-merges --format='%ad %ae %s'
2020-03-13 jmadill Add missing header to BUILD.gn.

Created with:
  roll-dep external/spirv-headers
2020-03-16 11:42:05 -04:00
Vasyl Teliman
7a8f79762c
Update dependencies (#3228) 2020-03-13 16:23:01 -04:00
greg-lunarg
1fe9bcc108
Instrument: Debug Printf support (#3215)
Create a pass to instrument OpDebugPrintf instructions.  This pass replaces all OpDebugPrintf instructions with instructions to write a record containing the string id and the all specified values into a special printf output buffer (if space allows). This pass is designed to support the printf validation in the Vulkan validation layers.

Fixes #3210
2020-03-12 09:19:52 -04:00