SEGV introduced between 20230804->20230808

[sthen]

Seen on OpenBSD -current amd64.

Seems there's a problem with the input buffer resizing done in ThomasDickey/mawk-snapshots@139ba08.

The initial report was from somebody using an ugly domain blocklist processor; here's a simpler reproducer:

$ yes | dd count=262144 bs=1 status=none > a
$ mawk '{ print $1 }' a
Segmentation fault (core dumped)

Program terminated with signal SIGSEGV, Segmentation fault.
#0  0x000006640d67210e in fillbuff (fd=3, target=0x6662dd36000 <error: Cannot access memory at address 0x6662dd36000>, size=0) at fin.c:413
413         *target = 0;
(gdb) bt
#0  0x000006640d67210e in fillbuff (fd=3, target=0x6662dd36000 <error: Cannot access memory at address 0x6662dd36000>, size=0) at fin.c:413
#1  0x000006640d6714e2 in FINgets (fin=0x666a8410258, len_p=0x772ef7dedfb8) at fin.c:246
#2  0x000006640d6584e2 in execute (cdp=0x666a840ac78, sp=0x6640d67b248 <execute.missing+8>, fp=0x0) at execute.c:1188                                                                                     #3  0x000006640d65411d in main (argc=4, argv=0x772ef7dee108)
at main.c:71

[sthen]

(Seems triggered by any file above 256k)

[ThomasDickey]

thanks - with valgrind I can see a problem which may be the same issue.

[ThomasDickey]

simple fix (which works for the test cases I have at hand):

--- fin.c	2023/08/08 21:57:08	1.51
+++ fin.c	2023/10/31 21:10:12
@@ -410,7 +410,6 @@
 	}
 
   out:
-    *target = 0;
     return (size_t) (entry_size - size);
 }

[sthen]

Thanks - this works for the other cases I've tried too.

[ThomasDickey]

I applied that on top of some other fixes, and will probably upload a new snapshot today or tomorrow.

[ThomasDickey]

done - https://invisible-island.net/mawk/CHANGES.html#t20231102