Null pointer when setting Mip LOD Bias

[kg]

states[i].value.values is NULL. It's a texture sampler. The name and semantic are also NULL. This is in EffectPass.Apply. The technique is ScreenSpaceBitmapWithLUTTechnique from https://github.com/sq/Fracture/blob/master/Squared/RenderLib/Shaders/SquaredBitmapShader.fx

[flibitijibibo]

From what I can tell your shader doesn't have any references to MIPMAPLODBIAS... my guess is these values are all just nonsense; I base this mostly on my attempt to parse a binary version of that shader:

[flibitijibibo@flibitTower flibitBuild]$ gdb ./testparse 
GNU gdb (GDB) Fedora 8.3-2.fc30
Copyright (C) 2019 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
Type "show copying" and "show warranty" for details.
This GDB was configured as "x86_64-redhat-linux-gnu".
Type "show configuration" for configuration details.
For bug reporting instructions, please see:
<http://www.gnu.org/software/gdb/bugs/>.
Find the GDB manual and other documentation resources online at:
    <http://www.gnu.org/software/gdb/documentation/>.

For help, type "help".
Type "apropos word" to search for commands related to "word"...
Reading symbols from ./testparse...
(gdb) run glsl120 bins/SquaredBitmapShader.fxb 
Starting program: /home/flibitijibibo/Programming/reProjects/mojoshader/flibitBuild/testparse glsl120 bins/SquaredBitmapShader.fxb
Missing separate debuginfos, use: dnf debuginfo-install glibc-2.29-12.fc30.x86_64
MojoShader testparse
Compiled against changeset hg-1202:3e07483a681c
Linked against changeset hg-1202:3e07483a681c

testparse: /home/flibitijibibo/Programming/reProjects/mojoshader/mojoshader_effects.c:375: readvalue: Assertion `mem->info.parameter_class >= MOJOSHADER_SYMCLASS_SCALAR && mem->info.parameter_class <= MOJOSHADER_SYMCLASS_VECTOR' failed.

Program received signal SIGABRT, Aborted.
0x00007ffff7cbfeb5 in raise () from /lib64/libc.so.6
(gdb) bt
#0  0x00007ffff7cbfeb5 in raise () from /lib64/libc.so.6
#1  0x00007ffff7caa895 in abort () from /lib64/libc.so.6
#2  0x00007ffff7caa769 in __assert_fail_base.cold () from /lib64/libc.so.6
#3  0x00007ffff7cb85b6 in __assert_fail () from /lib64/libc.so.6
#4  0x0000000000417d81 in readvalue (base=0x7ffff7b90018 "", typeoffset=4, valoffset=140, value=0x42b150, 
    objects=0x42a920, m=0x40c2d5 <MOJOSHADER_internal_malloc>, d=0x0)
    at /home/flibitijibibo/Programming/reProjects/mojoshader/mojoshader_effects.c:375
#5  0x0000000000418228 in readparameters (numparams=19, base=0x7ffff7b90018 "", ptr=0x7fffffffd210, 
    len=0x7fffffffd20c, params=0x42a8c0, objects=0x42a920, m=0x40c2d5 <MOJOSHADER_internal_malloc>, d=0x0)
    at /home/flibitijibibo/Programming/reProjects/mojoshader/mojoshader_effects.c:476
#6  0x00000000004197e6 in MOJOSHADER_parseEffect (profile=0x7fffffffd7e6 "glsl120", 
    buf=0x7ffff7b90010 "\001\t\377\376|\f", _len=41156, swiz=0x0, swizcount=0, smap=0x0, smapcount=0, 
    m=0x40c2d5 <MOJOSHADER_internal_malloc>, f=0x40c300 <MOJOSHADER_internal_free>, d=0x0)
    at /home/flibitijibibo/Programming/reProjects/mojoshader/mojoshader_effects.c:918
#7  0x0000000000403cdf in do_parse (fname=0x7fffffffd7ee "bins/SquaredBitmapShader.fxb", 
    buf=0x7ffff7b90010 "\001\t\377\376|\f", len=41156, prof=0x7fffffffd7e6 "glsl120")
    at /home/flibitijibibo/Programming/reProjects/mojoshader/utils/testparse.c:687
#8  0x0000000000403f10 in main (argc=3, argv=0x7fffffffd3f8)
    at /home/flibitijibibo/Programming/reProjects/mojoshader/utils/testparse.c:739
(gdb)

Looks like you've still got nested structs:

https://github.com/FNA-XNA/MojoShader/blob/fna/mojoshader_effects.c#L374

This is more of a MojoShader issue than FNA, so I've filed an issue over there for you: icculus/mojoshader#8

[flibitijibibo]

Oh, it's even better than that, it's actually a struct with multidimensional parameters, which apparently didn't have a pleasant layout last time I looked at it (float4x4 rather than float4). Still roughly the same area to look at for fixes.

[kg]

OK, I can try adjusting my shaders to see if that fixes this, or I can just see if I can fix this layout issue as well.

[flibitijibibo]

The hg diff in the MojoShader issue seems to produce the correct struct size, so as long as the matrices don't need to be transposed at CommitChanges for some reason (SetValueMatrix should be doing that, right?) then this will be a quicker fix than expected, at least for matrices.

[kg]

That's great. I don't actually see any nested structs in this shader's source so I bet it was just that diff.

[flibitijibibo]

Also discovered our LODBIAS friend in the headers...

https://github.com/sq/Fracture/blob/master/Squared/RenderLib/Shaders/BitmapCommon.fxh#L26

The parser appears to pull this in just fine:

    PARAM #7
        VALUE: TextureSampler -> (null)
            CLASS: OBJECT
            TYPE: SAMPLER
            ROWS/COLUMNS/ELEMENTS: 0, 0, 0
            TOTAL VALUES: 2
            SAMPLER VALUES:
                TYPE: TEXTURE -> 2
                TYPE: MIPMAPLODBIAS -> 0.00

Texture "2" being the object index for the "BitmapTexture" mapping and MIPMAPLODBIAS correctly printing the value (I threw in some arbitrary values and it was correct each time).

[kg]

That makes sense. If it's incompatible I can pull it out of the sampler and set it in code, anyway.

[flibitijibibo]

GUESS WHO'S FIRED TODAY

ᵐᵉᵉᵉᵉᵉᵉᵉᵉᵉᵉ

50bfe12