search

mm2 / Little-CMS

A free, open source, CMM engine. It provides fast transforms between ICC profiles.
https://www.littlecms.com
MIT License
549 stars 174 forks source link

Optimize cmsOpenIOhandlerFromMem() in 'w' mode #443

Closed x2018 closed 5 months ago

x2018 commented 5 months ago

Let the IOhandler successfully created by cmsOpenIOhandlerFromMem always has a valid Block.

mm2 commented 5 months ago

Thanks for your contribución. Just to clarify the issue you are solving... how you do to get "Block" member set to NULL?

x2018 commented 5 months ago

Thank you for your prompt reply.

From the view of lcms, it is meaningless to set Block to NULL. But for the users in practice, they may tranfer NULL due to unintentional reasons such as the memroy allocation problems.

After the changes, at least lcms will be more robust within its scope.

mm2 commented 5 months ago

Sorry, I didn't explain myself. I mean, which sequence of lcms functions you call to get "Block" set to NULL? I am asking because in the case it is not possible to set this value to NULL, then this check is useless, I.e, why not check if somebody has set it to 42 instead?

x2018 commented 5 months ago

I'm sorry that I misunderstood your meaning.

You are right, the first changed line of MemoryWrite() will indeed be theoretically useless, because the Block should not be NULL for a valid IOhandler after the second changed line takes effect.

Prior to this, as cmsOpenIOhandlerFromMem() itself is an exported function, the Block can be NULL if the Buffer passed to cmsOpenIOhandlerFromMem() is NULL in 'w' mode, since cmsOpenIOhandlerFromMem() only check it when it is 'r' mode without checking it in 'w' mode.

Because the test case crashed at memmove() in MemoryWrite(), I just made that change at the beginning. Now I have removed it.