Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Added ImageSourceData to TAGS_V2 #7053

Merged
merged 2 commits into from
Apr 1, 2023
Merged

Conversation

radarhere
Copy link
Member

In #7008, two images were reported to cause a RuntimeError. Testing on my machine, I found that this caused a segfault instead.

It turns out that this was due to an incorrect tag type being used for ImageSourceData.

Except then I found the RuntimeError, happening on 32-bit Windows.

E       RuntimeError: Error setting from dictionary

C:\hostedtoolcache\windows\Python\3.7.9\x86\lib\site-packages\pillow-9.5.0.dev0-py3.7-win32.egg\PIL\Image.py:437: RuntimeError
---------------------------- Captured stderr call -----------------------------
_TIFFVSetField: C:\Users\runneradmin\AppData\Local\Temp\pytest-of-runneradmin\pytest-0\test_save_imagesourcedata0\temp.tif: Bad LONG8 or IFD8 value 1000052142389717520 for "EXIFIFDOffset" tag 34665 in ClassicTIFF. Tag won't be written to file.

When it states 'Bad LONG8 or IFD8 value 1000052142389717520 for "EXIFIFDOffset" tag 34665', that number wasn't the value being passed through for the EXIF IFD offset from Python. When we are about to pass the value through to libtiff in C, we cast it to UINT32.

Pillow/src/encode.c

Lines 906 to 907 in 3166901

status = ImagingLibTiffSetField(
&encoder->state, (ttag_t)key_int, (UINT32)PyLong_AsLong(value));

When libtiff is throwing the error, the variable is uint64_t. Changing UINT32 to uint64_t fixed the RunTimeError.

@radarhere radarhere added the TIFF label Mar 31, 2023
src/encode.c Outdated
@@ -904,7 +904,7 @@ PyImaging_LibTiffEncoderNew(PyObject *self, PyObject *args) {
&encoder->state, (ttag_t)key_int, (UINT16)PyLong_AsLong(value));
} else if (type == TIFF_LONG) {
status = ImagingLibTiffSetField(
&encoder->state, (ttag_t)key_int, (UINT32)PyLong_AsLong(value));
&encoder->state, (ttag_t)key_int, (uint64_t)PyLong_AsLong(value));
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't fully understand this change and can't look into it right now (I'm not sure why the TIFF_LONG8 error was triggered when this change is for TIFF_LONG), but at least on Windows, the PyLong_AsLong function returns the equivalent of int32_t. I think PyLong_AsLongLong should return int64_t on all platforms, although I'm not sure if it matters here.

To be clear, using PyLong_AsLong here instead of PyLong_AsLongLong is not a security issue, I'm just not sure whether it could save the wrong values.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sure, that seems better than casting the value. I've updated the commit.

Copy link
Contributor

@nulano nulano Apr 1, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

PyLong_AsLongLong returns int64_t, not uint64_t. It probably doesn't matter due to how signedness casting works, but it's not the same type. I was just pointing out that PyLong_AsLong has a different output range on Windows vs Linux.

@mgorny
Copy link
Contributor

mgorny commented Nov 2, 2024

This change broke Pillow on PowerPC: #8522

To be honest, I don't really understand what's being done here. Given that TIFF_LONG is defined as "32-bit unsigned integer", passing a 64-bit integer (signed or unsigned) seems wrong. I'm guessing on other architectures VA are passed in such a way as to conceal the problem, and it explodes on PowerPC.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants