Age | Commit message (Collapse) | Author |
|
Do this by making CFX_CodecMemory actually own the memory that
it is ref-counting. Remove some test cases that are now prohibited,
and relax one lifetime restriction in the test because we are now
doing one additional copy (in the test, but not in real life).
Bug:879512
Change-Id: If030dfcf97fe40155c46a42288fc73192437ce9c
Reviewed-on: https://pdfium-review.googlesource.com/c/43670
Commit-Queue: Tom Sepez <tsepez@chromium.org>
Reviewed-by: Lei Zhang <thestig@chromium.org>
|
|
Next step towards making CFX_CodecMemory own the memory it is
ref-counting.
Change-Id: I5922d80d13032c3ea028447c47d34a10234109a3
Reviewed-on: https://pdfium-review.googlesource.com/c/43630
Reviewed-by: Lei Zhang <thestig@chromium.org>
Commit-Queue: Tom Sepez <tsepez@chromium.org>
|
|
Change-Id: I92aa74c86aef890fcb991f039b9707d680fdb83c
Reviewed-on: https://pdfium-review.googlesource.com/c/43652
Reviewed-by: Lei Zhang <thestig@chromium.org>
Commit-Queue: Tom Sepez <tsepez@chromium.org>
|
|
- Fix buffer leak if TryRealloc fails.
- Make m_SrcSize always represent the usable bytes in the buffer, even
when read the last partial block from the file.
- Remove redundant comparisons and use std::min().
- Better naming.
Change-Id: Ie7dd79bac21b2f2422f299563a2dd28ed358e3e2
Reviewed-on: https://pdfium-review.googlesource.com/42130
Commit-Queue: Tom Sepez <tsepez@chromium.org>
Reviewed-by: Lei Zhang <thestig@chromium.org>
|
|
Extract some dividends from having landed the previous CL.
Change-Id: I9ade77c801b51762cab801716b9fd2bc4c634036
Reviewed-on: https://pdfium-review.googlesource.com/c/41990
Commit-Queue: Tom Sepez <tsepez@chromium.org>
Reviewed-by: Lei Zhang <thestig@chromium.org>
|
|
Because I nearly botched this trivial calculation in the previous CL.
Change-Id: I7438f9d3476d93b7899c2d7d761234769f53f9e3
Reviewed-on: https://pdfium-review.googlesource.com/43010
Commit-Queue: Lei Zhang <thestig@chromium.org>
Reviewed-by: Lei Zhang <thestig@chromium.org>
|
|
Another step before trying to fix the memory issue. Forces common
APIs on the bunch of decoders, though some methods are unused.
Requires adding some arguments/return values to get to a common
API which are not used in all cases (yet?). Required converting
some args to spans. Required proxying a GetJumpMark() call through
the public module API to the private context.
Bug: pdfium:1082
Change-Id: I0c0b7415141ff2a6f4f44777ca3d05521f08130d
Reviewed-on: https://pdfium-review.googlesource.com/41950
Commit-Queue: Lei Zhang <thestig@chromium.org>
Reviewed-by: Lei Zhang <thestig@chromium.org>
|
|
Small consolidation before trying to fix a memory issue.
Unfortunately, this involves converting a switch -> ifs, since we
no longer dispatch exactly the same.
Change-Id: I7ef61db98cdfbbc983adf21e7b3fe4ef1d2ce869
Reviewed-on: https://pdfium-review.googlesource.com/41830
Reviewed-by: Lei Zhang <thestig@chromium.org>
Commit-Queue: Tom Sepez <tsepez@chromium.org>
|
|
It's always passed as nullptr. It's a good thing, too, since it looks
like some of the implementations are returning a wrong pointer.
Add some missing |const|s.
Change-Id: I768048fdfe4cdd1dc838fee26fec18e024e39920
Reviewed-on: https://pdfium-review.googlesource.com/41810
Reviewed-by: Lei Zhang <thestig@chromium.org>
Commit-Queue: Tom Sepez <tsepez@chromium.org>
|
|
The associated bug is caused by failure to clean up context before
returing at line 740 or so. But cleanups are fragile, so instead
do not move context out of a local until success. This in turn
means passing the local to one helper rather that using the stored
value and confirming this is the only helper method invoked during
the time in which the context is not yet stored.
Bug: 879025
Change-Id: I1acb438edc28f7f4dd8acccc2461fe3a5ea7f5c8
Reviewed-on: https://pdfium-review.googlesource.com/41670
Reviewed-by: Lei Zhang <thestig@chromium.org>
Commit-Queue: Tom Sepez <tsepez@chromium.org>
|
|
This requires making some members unique_ptr<> to preserve the
destruction order.
Bug: 877766
Change-Id: I69e0173fdde7e9cd0d24017d6c829b9b9cf600c0
Reviewed-on: https://pdfium-review.googlesource.com/41490
Commit-Queue: Tom Sepez <tsepez@chromium.org>
Reviewed-by: Lei Zhang <thestig@chromium.org>
|
|
Rename .cpp file to match class and .h file (!!!)
Put initialziations into header.
Change-Id: I2cbc7608287f8df887ce186d122dcd3db2aef19b
Reviewed-on: https://pdfium-review.googlesource.com/41511
Commit-Queue: Tom Sepez <tsepez@chromium.org>
Reviewed-by: Lei Zhang <thestig@chromium.org>
|