I see an assert() being triggered in debug mode:
mutool: source/fitz/colorspace.c:3245: std_conv_color: Assertion `srcs->to_ccs && dsts->from_ccs' failed.

This is because both srcs->to_ccs and dsts->from_css are == NULL. We're attempting to a conversion CalGray -> CalRGB. Both pointers are set to NULL when fz_new_cal_colorspace() sets both conversion functions to NULL. If there is no CMS then k_default_gray and k_default_rgb set valid pointers.

Looks like fz_new_cal_colorspace() really ought to set the pointers..?

Perhaps fz_find_color_converter() ought to handle Cal-colorspaces like it handles ICC-colorspaces since fz_source_colorspace_cm() seems to look for both types, but in the succeeding test Cal-colorspaces are not checked for.

I tried to do:

diff --git a/source/fitz/colorspace.c b/source/fitz/colorspace.c
index 336513fe..45cd59e4 100644
--- a/source/fitz/colorspace.c
+++ b/source/fitz/colorspace.c
@@ -3418,7 +3418,7 @@ void fz_find_color_converter(fz_context *ctx, fz_color_converter *cc, const fz_c
        else
        {
                const fz_colorspace *ss_base = fz_source_colorspace_cm(ctx, ss);
-               if (ss_base != NULL && fz_colorspace_is_icc(ctx, ds))
+               if (ss_base != NULL && (fz_colorspace_is_icc(ctx, ds) || fz_colorspace_is_cal(ctx, ds)))
                {
                        if (ss_base == ss)
                                cc->convert = icc_conv_color;


But that lead to a crash inside cmsGetColorSpace() later on, presumably because fz_cmm_init_profile() failed to make a profile in fz_get_icc_link() and then went on to use the default.

I'm out of my depth here. Either someone with more knowledge needs to look into this one or I have to spend more time trying to figure out how this is supposed to work.

I don't sure it's affect other pdfs, but this line fix this file


- if (srcs != dsts)
+ if (srcs != dsts && srcs->to_ccs != NULL)

(In reply to foobnix from comment #3)
> I don't sure it's affect other pdfs, but this line fix this file
> 
> 
> - if (srcs != dsts)
> + if (srcs != dsts && srcs->to_ccs != NULL)

Yes I can see that this would help, but then the conversion from CalGray to CalRGB might not be correct.

I see the issue.  I should have a fix for this shortly.

This works fine with head.  It think it was fixed with 

http://git.ghostscript.com/?p=mupdf.git;a=commit;h=56a752c928146b94838b09c7269c50b0aa800e68