Quoting Phil Longstaff <[EMAIL PROTECTED]>:
I've been playing around with G2 and valgrind 3.1.0 and have seen the
following issues on SuSE 9.3:
1) There are lots of messages about decisions based on uninitialized
variables
and accessing past the end of malloc'ed mem in the code dealing with guids.
It seems as though readdir() does not completely initialize the dirent struct
that it returns. There were also messages that md5_process_bytes() was
accessing bytes from the dirent that were off the end of memory malloc'ed by
opendir().
These are all perfectly fine. It's just trying to get random data, so
the fact that the data is uninitialized is okay. If it bothers you,
you should add a valgrind supression.
Finally, there were messages that guid_equal() was returning a
result based on uninitialized values because the comparison was done
including the __align_me field, not just the guid data.
This should get fixed.
The attached patch replaces the const 16 with GUID_DATA_SIZE, only compares
the guid data, not the other field in the struct, and removes the use of the
dirent structure in md5 calculation.
The first part of the patch is okay in theory (I haven't looked at the code).
The second part of the patch is not okay.
2) After g2 is up, I open an account, switch to ledger view, put my cursor
into a split and start pressing delete-tab-delete-tab... I then get the
listing shown in attachment gnucash.valgrind. I added printf statements to
show as splits were allocated and freed. Most serious is that a split
continues to be accessed after it is freed.
This is certainly something that needs to get fixed! Thank you. Hopefully
someone could track down the reference?
Phil
-derek
--
Derek Atkins, SB '93 MIT EE, SM '95 MIT Media Laboratory
Member, MIT Student Information Processing Board (SIPB)
URL: http://web.mit.edu/warlord/ PP-ASEL-IA N1NWH
[EMAIL PROTECTED] PGP key available
_______________________________________________
gnucash-devel mailing list
[email protected]
https://lists.gnucash.org/mailman/listinfo/gnucash-devel
