diff --git a/cmake/compile-flags.cmake b/cmake/compile-flags.cmake
index c31ba18..078c42e 100644
--- a/cmake/compile-flags.cmake
+++ b/cmake/compile-flags.cmake
@@ -19,7 +19,6 @@ add_compile_flags(BOTH
         -ffunction-sections
         -fdata-sections
         -fstrict-volatile-bitfields
-        -fno-zero-initialized-in-bss
         -O2
         -ggdb
         )
diff --git a/lds/kendryte.ld b/lds/kendryte.ld
index fad6b3f..20c6869 100644
--- a/lds/kendryte.ld
+++ b/lds/kendryte.ld
@@ -53,12 +53,14 @@ SECTIONS
     KEEP( *(.text.start) )
     KEEP( *(.text.systick) )
   } > ram : DATA
-
+  
   .init :
   {
     KEEP (*(SORT_NONE(.init)))
   } > ram : DATA
-
+  
+  . = ALIGN(8);
+  . = .;
   .text :
   {
     *(.text.unlikely .text.*_unlikely .text.unlikely.*)
@@ -163,6 +165,7 @@ SECTIONS
   } > ram : DATA
   _edata = .; PROVIDE (edata = .);
   
+  . = ALIGN(8);
   . = .;
   __bss_start = .;
   .sbss2 : { *(.sbss2 .sbss2.* .gnu.linkonce.sb2.*) } > ram : DYN_DATA
diff --git a/lib/bsp/crt.S b/lib/bsp/crt.S
index bfa8938..12525cc 100644
--- a/lib/bsp/crt.S
+++ b/lib/bsp/crt.S
@@ -123,6 +123,15 @@ _start:
   jr t1
 2:
   la sp, _idle_stack0_top
+  
+  # clear the bss segment
+  la t0, __bss_start
+  la t1, __bss_end
+1:
+  sd zero, 0(t0)
+  addi t0, t0, 8
+  bltu t0, t1, 1b
+
   la t0, _init_bsp
   jr t0
   
diff --git a/lib/bsp/entry_user.c b/lib/bsp/entry_user.c
index 355fe71..134b9d6 100644
--- a/lib/bsp/entry_user.c
+++ b/lib/bsp/entry_user.c
@@ -27,8 +27,6 @@
 #define PLL1_OUTPUT_FREQ 160000000UL
 #define PLL2_OUTPUT_FREQ 45158400UL
 
-extern uint8_t __bss_start[];
-extern uint8_t __bss_end[];
 extern uint8_t _tls_data[];
 
 extern int main(int argc, char *argv[]);
@@ -40,15 +38,8 @@ static void setup_clocks()
     sysctl_pll_set_freq(SYSCTL_PLL2, PLL2_OUTPUT_FREQ);
 }
 
-static void init_bss(void)
-{
-    memset(__bss_start, 0, __bss_end - __bss_start);
-}
-
 void _init_bsp()
 {
-    /* Initialize bss data to 0 */
-    init_bss();
     /* Init FPIOA */
     fpioa_init();
     bsp_pin_setup();
diff --git a/lib/bsp/except.c b/lib/bsp/except.c
index 1a0f6fb..33dd771 100644
--- a/lib/bsp/except.c
+++ b/lib/bsp/except.c
@@ -20,49 +20,49 @@
 void __attribute__((weak)) handle_misaligned_fetch(uintptr_t *regs, uintptr_t cause)
 {
     dump_core("misaligned fetch", regs, cause);
-    exit(1337);
+    sys_exit(1337);
 }
 
 void __attribute__((weak)) handle_fault_fetch(uintptr_t *regs, uintptr_t cause)
 {
     dump_core("fault fetch", regs, cause);
-    exit(1337);
+    sys_exit(1337);
 }
 
 void __attribute__((weak)) handle_illegal_instruction(uintptr_t *regs, uintptr_t cause)
 {
     dump_core("illegal instruction", regs, cause);
-    exit(1337);
+    sys_exit(1337);
 }
 
 void __attribute__((weak)) handle_breakpoint(uintptr_t *regs, uintptr_t cause)
 {
     dump_core("breakpoint", regs, cause);
-    exit(1337);
+    sys_exit(1337);
 }
 
 void __attribute__((weak)) handle_misaligned_load(uintptr_t *regs, uintptr_t cause)
 {
     dump_core("misaligned load", regs, cause);
-    exit(1337);
+    sys_exit(1337);
 }
 
 void __attribute__((weak)) handle_fault_load(uintptr_t *regs, uintptr_t cause)
 {
     dump_core("fault load", regs, cause);
-    exit(1337);
+    sys_exit(1337);
 }
 
 void __attribute__((weak)) handle_misaligned_store(uintptr_t *regs, uintptr_t cause)
 {
     dump_core("misaligned store", regs, cause);
-    exit(1337);
+    sys_exit(1337);
 }
 
 void __attribute__((weak)) handle_fault_store(uintptr_t *regs, uintptr_t cause)
 {
     dump_core("fault store", regs, cause);
-    exit(1337);
+    sys_exit(1337);
 }
 
 void handle_except(uintptr_t *regs, uintptr_t cause)
diff --git a/lib/bsp/include/interrupt.h b/lib/bsp/include/interrupt.h
index 259a17f..0727e34 100644
--- a/lib/bsp/include/interrupt.h
+++ b/lib/bsp/include/interrupt.h
@@ -46,6 +46,7 @@ void handle_irq_m_soft(uintptr_t *regs, uintptr_t cause);
 void handle_irq_m_timer(uintptr_t *regs, uintptr_t cause);
 void handle_irq_m_ext(uintptr_t *regs, uintptr_t cause);
 void handle_syscall(uintptr_t *regs, uintptr_t cause);
+void __attribute__((noreturn)) sys_exit(int code);
 #ifdef __cplusplus
 }
 #endif
diff --git a/lib/freertos/conf/FreeRTOSConfig.h b/lib/freertos/conf/FreeRTOSConfig.h
index 9d78392..5fb51fd 100644
--- a/lib/freertos/conf/FreeRTOSConfig.h
+++ b/lib/freertos/conf/FreeRTOSConfig.h
@@ -92,7 +92,7 @@ enum
 #define configUSE_DAEMON_TASK_STARTUP_HOOK		0
 
 /* memory */
-#define configMINIMAL_STACK_SIZE			( ( unsigned short ) 1024 )
+#define configMINIMAL_STACK_SIZE			( ( unsigned short ) 20480 )
 #define configTOTAL_HEAP_SIZE				( ( size_t ) ( 1024 * 1024 ) )
 #define configSUPPORT_STATIC_ALLOCATION			1
 #define configSUPPORT_DYNAMIC_ALLOCATION		1
diff --git a/lib/freertos/os_entry.c b/lib/freertos/os_entry.c
index b818be7..86ec95c 100644
--- a/lib/freertos/os_entry.c
+++ b/lib/freertos/os_entry.c
@@ -48,10 +48,6 @@ static void main_thunk(void *p)
     /* Init libc array for C++ */
     __libc_init_array();
 
-    clear_csr(mie, MIP_MTIP);
-    clint_ipi_enable();
-    set_csr(mstatus, MSTATUS_MIE);
-
     install_hal();
     install_drivers();
     configure_fpioa();
@@ -71,6 +67,10 @@ static void os_entry_core1()
 
 int os_entry(int (*user_main)(int, char **))
 {
+    clear_csr(mie, MIP_MTIP);
+    clint_ipi_enable();
+    set_csr(mstatus, MSTATUS_MIE);
+
     TaskHandle_t mainTask;
     main_thunk_param_t param = {};
     param.user_main = user_main;
diff --git a/lib/freertos/portable/port.c b/lib/freertos/portable/port.c
index c132e90..f93e081 100644
--- a/lib/freertos/portable/port.c
+++ b/lib/freertos/portable/port.c
@@ -160,7 +160,7 @@ int vPortSetInterruptMask(void)
     int ret;
     __asm volatile("csrr %0,mie"
                    : "=r"(ret));
-    __asm volatile("csrc mie,%0" ::"i"(7));
+    __asm volatile("csrc mie,%0" ::"r"(0x888));
     return ret;
 }
 /*-----------------------------------------------------------*/