Keep having an error when using static array which is not initialized

I am using 16bit microcontroller which uses gcc4.5 compiler, 32kbyte flash/1kbyte ram. compiling is fine without error, memory is reaching to it's limit but it has some vacancy on both flash/ram and works just fine

However, when i define static array without initialization it causes system to die. I cannot use debugger to watch what's going on inside my system.

I didn't even use the array other than define it. so I assume compiler should skip below code but it kept making a problem.

typedef enum
{
    PTC_VOLT_ERROR = 0,
    PTC_LINE_ERROR,
    PTC_ERROR,  
    PTC_PCB1TEMP_ERROR,
    PTC_PCB2TEMP_ERROR, 
    PTC_COMM_ERROR,
    PTC_SELF_DIAG_ERROR,
    PTC_PCB1TEMPSSR_ERROR,
    PTC_PCB2TEMPSSR_ERROR,
    NUM_OF_PTC_ERROR_TYPE
}PtcErrorTypeInfo_t;
(...)
static INT8U preErrorInfo[NUM_OF_PTC_ERROR_TYPE]; // This cause system to die
//static INT8U preErrorInfo[NUM_OF_PTC_ERROR_TYPE]={0,}; // This is ok

Then I tried with non-static global arry and found it was working well.

INT8U preErrorInfo[NUM_OF_PTC_ERROR_TYPE]; // This is ok
static INT8U preErrorInfo[2]; // This is also ok


static INT8U preErrorInfo[9]; // This is not working
static INT8U preErrorInfo[2]; // This is working

I want to know what could make my system down?

→→→ It turned out that it had something to do with overflow and trap error.

first>> when I compile without static, in the map file

 .data.g_PtcProtectionState
                0x0000007e        0x2 obj/EcuProtectCTR.o
                0x0000007e                _g_PtcProtectionState
 .data.preErrorInfo
                0x00000080        0x9 obj/EcuProtectCTR.o
 .data.u1_ErrorDeterTmr
                0x00000089        0x1 obj/EcuProtectCTR.o
 .data.pendingCalibOn
                0x0000008a        0x6 obj/EcuSsrInputCTR.o
                0x0000008a                _pendingCalibOn

(...)
 .bss.g_u4_ErrorCntrInfo
                0x00000144        0x4 obj/EcuProtectCTR.o
 **.bss.g_u1_TmrInt
                0x00000148        0x1 obj/SchM.o
                0x00000148                _g_u1_TmrInt**
 .bss.u1_ElapsedTime
                0x00000149        0x1 obj/SchM.o

it makes preErrorInfo[9] array seats in the data area.

second>> if i compile with 'static'

 .bss.g_u4_ErrorCntrInfo
                0x0000013c        0x4 obj/EcuProtectCTR.o
 .bss.preErrorInfo
                0x00000140        0x9 obj/EcuProtectCTR.o
 .bss.g_u1_TmrInt
                0x00000149        0x1 obj/SchM.o
                0x00000149                _g_u1_TmrInt

preErrorInfo arry nicely seats on bss area. However, there is a variable 'g_u1_TmrInt' which keeps causing a problem. this timer-interrupt geard variable keeps making an overflow.

since 'static preErrorInfo array' will seat right above 'g_u1_TmrInt' somehow it could be messed up with 'static preErrorInfo array' when overflowing.

But I wonder why it didn't cause a problem when it comes to a case 'first>>' which has a 'g_u4_ErrorCntrInfo' 4byte array seated right above 'g_u1_TmrInt' and it would surely make the same overflow error.

※ when I change 'g_u1_TmrInt' type to 16bit(no-overflow and no error), it would look like this and work like charm.

 .bss.preErrorInfo
                0x00000140        0x9 obj/EcuProtectCTR.o
 *fill*         0x00000149        0x1 
 .bss.g_u1_TmrInt
                0x0000014a        0x2 obj/SchM.o
                0x0000014a                _g_u1_TmrInt
 .bss.u1_ElapsedTime
                0x0000014c        0x2 obj/SchM.o
 .bss.SchConfig