#1349 Crash when compiling array of negative size

[Patryk]

Compiler crashes after unusually long compile time (few minutes on AMD Athlon XP 1700+, WXP HE PL SP2).

Source:
xdata char tbl[-5] = {10};

command:
sdcc -c Source5.c

SDCC version:
SDCC : mcs51/gbz80/z80/avr/ds390/pic16/pic14/TININative/xa51/ds400/hc08
2.7.2 #4852 (Jun 15 2007) (MINGW32)

Error message:
Assertion failed: dbuf->buf != NULL, file /home/sdcc-builder/build/sdcc-build/or
ig/sdcc/src/../support/Util/dbuf_string.c, line 153

This application has requested the Runtime to terminate it in an unusual way.
Please contact the application's support team for more information.
Caught signal 22: SIGABRT

Expected behaviour: report error (array size negative).

[Borut Ražem]

Logged In: YES
user_id=568035
Originator: NO

Fixed in svn revision #4867.

Borut

[Patryk]

Logged In: YES
user_id=1788180
Originator: YES

Great!
BTW: you mean #4862?

[Patryk]

Logged In: YES
user_id=1788180
Originator: YES

Huh? Just found #4867 snapshot released. #4862 was latest just few minutes ago. What time snapshots are released? Or some browser cache issues?

[Borut Ražem]

Logged In: YES
user_id=568035
Originator: NO

Svn revision number is changed on every commit. There might be several commits between two snapshot builds, so don't assume:

new_snapshot_build_number) == old_snapshot_build_number + 1

but the following is always true:

new_snapshot_build_number > old_snapshot_build_number

Svn revision number is known at the commit time, which might be quite some time before the next snapshot build is available.

Borut

[Patryk]

Logged In: YES
user_id=1788180
Originator: YES

I supposed it is like you wrote, but isn't new snapshot builds available around midnight? I.e. #4862 contains all updates committed during June 27, and #4867 all during June 28, so today (June 28) #4862 is the latest and contain your fix done June 27. Apparently not since #4867 became available today around noon.

[Borut Ražem]

Logged In: YES
user_id=568035
Originator: NO

What do you mean with "midnight"? At which part of the world?
That's why I renamed "Nightly snapshot builds" to just "Snapshot builds" ;-)

Linux (i386 and amd64) and Windows builds start at 7h GMT and are available at cca. 8h GMT, Mac OS X build starts at 8h GMT and is available at cca. 9h GMT.

Borut

[Patryk]

Logged In: YES
user_id=1788180
Originator: YES

I see. I meant midnight of GMT. Anyway #4867 wasn't available around 09:35 GMT, but it must been some browser or server cache issue.

Now I can use:
typedef char TCompile_time_check[(sizeof(buf) >= sizeof(the_largest_object)) ? 1 : -1];
for some code sanity checks.

Or more convenient with macro:
#define GLUE_B(x, y) x##y
#define GLUE(x, y) GLUE_B(x, y)
#define COMPILE_TIME_CHECK(expr) typedef char GLUE(TCTChk_, __LINE__)[(expr) ? 1 : -1]

COMPILE_TIME_CHECK(sizeof(buf) >= sizeof(the_largest_object));

#define MAX_U_VAL(x) ( UINTMAX_MAX >> ((sizeof(UINTMAX_MAX) - sizeof(x))*CHAR_BIT) )
COMPILE_TIME_CHECK( MAX_U_VAL(table_index) >= (sizeof(table)/sizeof(table[0])) );

[Borut Ražem]

Logged In: YES
user_id=568035
Originator: NO

I made some mistakes in my previous posting:

- Mac OS X build starts at 7h GMT and is available at cca. 8h GMT,
- Linux (i386 and amd64) and Windows builds start at 8h GMT and are available at cca. 9:45h GMT

Borut

[Patryk]

Logged In: YES
user_id=1788180
Originator: YES

That means there were no browser or server cache issues :-)