FreeCalypso > hg > fc-tourmaline
view src/gpf/osl/os_mem_fl.c @ 303:f76436d19a7a default tip
!GPRS config: fix long-standing AT+COPS chance hanging bug
There has been a long-standing bug in FreeCalypso going back years:
sometimes in the AT command bring-up sequence of an ACI-only MS,
the AT+COPS command would produce only a power scan followed by
cessation of protocol stack activity (only L1 ADC traces), instead
of the expected network search sequence. This behaviour was seen
in different FC firmware versions going back to Citrine, and seemed
to follow some law of chance, not reliably repeatable.
This bug has been tracked down and found to be specific to !GPRS
configuration, stemming from our TCS2/TCS3 hybrid and reconstruction
of !GPRS support that was bitrotten in TCS3.2/LoCosto version.
ACI module psa_mms.c, needed only for !GPRS, was missing in the TCS3
version and had to be pulled from TCS2 - but as it turns out,
there is a new field in the MMR_REG_REQ primitive that needs to be
set correctly, and that psa_mms.c module is the place where this
initialization needed to be added.
author | Mychaela Falconia <falcon@freecalypso.org> |
---|---|
date | Thu, 08 Jun 2023 08:23:37 +0000 |
parents | 4e78acac3d88 |
children |
line wrap: on
line source
/* * This C module is a reconstruction based on the disassembly of * os_mem.obj in frame_na7_db_fl.lib from the Leonardo package. */ /* set of included headers from COFF symtab: */ #include <stdio.h> #include <string.h> #include "nucleus.h" #include "typedefs.h" #include "os.h" #include "gdi.h" #include "os_types.h" #include "os_glob.h" extern T_OS_PART_GRP_TABLE_ENTRY PartGrpTable[]; extern T_OS_MEM_POOL_TABLE_ENTRY MemPoolTable[]; extern T_OS_POOL_BORDER PoolBorder[]; OS_HANDLE os_ext_pool_handle; OS_HANDLE os_int_pool_handle; static USHORT NumOfMemoryPools; static NU_SEMAPHORE MemSemCB; static NU_MEMORY_POOL mem_pool_head; GLOBAL LONG os_SetPoolHandles(OS_HANDLE ext_pool_handle, OS_HANDLE int_pool_handle) { os_ext_pool_handle = ext_pool_handle; os_int_pool_handle = int_pool_handle; return(OS_OK); } static int os_GetPartitionPoolEntry(USHORT Index, T_OS_PART_POOL **pool) { static T_OS_PART_POOL *part_pool; static int grp_hndl; switch (Index) { case FIRST_ENTRY: grp_hndl = 0; *pool = part_pool = PartGrpTable[0].grp_head; return(OS_OK); case NEXT_ENTRY: if (part_pool->next) { *pool = part_pool = part_pool->next; return(OS_OK); } grp_hndl++; if (PartGrpTable[grp_hndl].grp_head) { *pool = part_pool = PartGrpTable[grp_hndl].grp_head; return(OS_OK); } else return(OS_ERROR); default: return(OS_ERROR); } } GLOBAL LONG os_PartitionInformation(USHORT Handle, char *Buffer) { T_OS_PART_POOL *pool; OPTION SuspendType; UNSIGNED PoolSize; UNSIGNED PartitionSize; UNSIGNED Available; UNSIGNED Waiting; UNSIGNED Allocated; VOID *pStartAddress; NU_TASK *First; CHAR Name[NU_MAX_NAME]; if (os_GetPartitionPoolEntry(Handle, &pool) == OS_ERROR) return(OS_ERROR); if (NU_Partition_Pool_Information(&pool->pcb, Name, &pStartAddress, &PoolSize, &PartitionSize, &Available, &Allocated, &SuspendType, &Waiting, &First) != NU_SUCCESS) return(OS_ERROR); sprintf(Buffer, "Name:%s Addr:%lx PoolSize:%ld PartSize:%ld Free:%ld Used:%ld", Name, (UNSIGNED) pStartAddress, PoolSize, PartitionSize, Available, Allocated); return(OS_OK); } static int os_GetMemoryPoolEntry(USHORT Index, OS_HANDLE *Handle) { static USHORT Idx; switch (Index) { case FIRST_ENTRY: Idx = 0; break; case NEXT_ENTRY: Idx++; break; default: Idx = Index; } if (Idx == NumOfMemoryPools) return(OS_ERROR); *Handle = Idx; return(OS_OK); } GLOBAL LONG os_MemoryInformation(USHORT Index, char *Buffer) { OS_HANDLE Handle; OPTION SuspendType; UNSIGNED Size, Min, Available, Waiting; VOID *pStartAddress; NU_TASK *First; CHAR Name[NU_MAX_NAME]; if (os_GetMemoryPoolEntry(Index, &Handle) == OS_ERROR) return(OS_ERROR); if (NU_Memory_Pool_Information(MemPoolTable[Handle].pcb, Name, &pStartAddress, &Size, &Min, &Available, &SuspendType, &Waiting, &First) != NU_SUCCESS) return(OS_ERROR); sprintf(Buffer, "Heapname:%s Addr:%lx Size:%ld Min:%ld Free:%ld Suspend:%d", Name, (UNSIGNED) pStartAddress, Size, Min, Available, SuspendType); return(OS_OK); } GLOBAL LONG os_MemInit(void) { USHORT i; if (NU_Create_Semaphore(&MemSemCB, "MEMSEM", 1, NU_PRIORITY) != NU_SUCCESS) return(OS_ERROR); for (i = 0; i <= MaxPoolGroups; i++) { PoolBorder[i].Start = (char *)0xFFFFFFFF; PoolBorder[i].End = (char *)0; PartGrpTable[i].grp_head = 0; PartGrpTable[i].name[0] = 0; } MemPoolTable[0].pcb = &mem_pool_head; return(OS_OK); } void os_InitPartitionCheck(T_OS_PART_POOL *pool) { unsigned **Buffer, offset; USHORT i, k; NU_Allocate_Memory(MemPoolTable[0].pcb, (VOID **) &Buffer, pool->pcb.pm_available * sizeof(unsigned *), NU_NO_SUSPEND); offset = pool->pcb.pm_partition_size / sizeof(unsigned) - 1; for (i = 0; ; i++) { if (NU_Allocate_Partition(&pool->pcb, (VOID **)(Buffer + i), NU_NO_SUSPEND) != NU_SUCCESS) break; Buffer[i][offset] = GUARD_PATTERN; } for (k = 0; k < i; k++) if (NU_Deallocate_Partition(Buffer[k]) != NU_SUCCESS) break; NU_Deallocate_Memory(Buffer); } GLOBAL const ULONG * os_GetPrimpoolCB(int grp, int id) { T_OS_PART_POOL *pool; int i; pool = PartGrpTable[grp].grp_head; if (!pool) return(0); if (id < 0) return(0); for (i = 0; i < id; i++) { pool = pool->next; if (!pool) return(0); } return (const ULONG *) &pool->pcb; } GLOBAL LONG os_GetPartitionPoolStatus(ULONG size, OS_HANDLE gr_hndl, USHORT *m_free, USHORT *m_alloc) { T_OS_PART_POOL *pool; UNSIGNED dummy, allocated, available; CHAR Name[NU_MAX_NAME]; for (pool = PartGrpTable[gr_hndl].grp_head; pool; pool = pool->next) { if (!size) break; if (size > pool->size) continue; if (NU_Partition_Pool_Information(&pool->pcb, Name, (VOID **)&dummy, &dummy, &dummy, &available, &allocated, (OPTION *)&dummy, &dummy, (NU_TASK **)&dummy) != NU_SUCCESS) break; *m_alloc = allocated; *m_free = available; return(OS_OK); } *m_alloc = 0; *m_free = 0; return(OS_ERROR); } GLOBAL LONG os_GetPartitionGroupHandle(OS_HANDLE Caller, char *Name, OS_HANDLE *GroupHandle) { int i; for (i = 0; i <= MaxPoolGroups; i++) { if (!PartGrpTable[i].grp_head) continue; if (strncmp(Name, PartGrpTable[i].name, RESOURCE_NAMELEN-1)) continue; *GroupHandle = i; return(OS_OK); } return(OS_ERROR); } GLOBAL LONG os_DeallocateMemory(OS_HANDLE TaskHandle, T_VOID_STRUCT *Buffer) { if (NU_Deallocate_Memory(Buffer) == NU_SUCCESS) return(OS_OK); else return(OS_ERROR); } GLOBAL LONG os_AllocateMemory(OS_HANDLE TaskHandle, T_VOID_STRUCT **Buffer, ULONG Size, ULONG Suspend, OS_HANDLE PoolHandle) { int ret, sts; if (Suspend == 0xFFFFFFFF) Suspend = 1; ret = OS_OK; for (;;) { sts = NU_Allocate_Memory(MemPoolTable[PoolHandle].pcb, (VOID **) Buffer, Size, Suspend); switch (sts) { case NU_SUCCESS: return(ret); case NU_INVALID_SUSPEND: Suspend = 0; continue; case NU_NO_MEMORY: case NU_TIMEOUT: if (Suspend == 1) { Suspend = 0xFFFFFFFF; ret = OS_WAITED; continue; } else { *Buffer = 0; return(OS_TIMEOUT); } default: /* * Disassembly reveals that the original code * has an endless loop here, the equivalent * of continue. My guess is that they simply * forgot the default case, and so control * falls onto the closing brace of the switch * and then onto the closing brace of the for * loop. But I prefer better error handling, * hence the present addition. - Space Falcon */ *Buffer = 0; return(OS_ERROR); } } } GLOBAL LONG os_CreatePartitionPool(OS_HANDLE TaskHandle, char *GroupName, void *Addr, USHORT Num, ULONG Size, OS_HANDLE *GroupHandle) { STATUS sts; T_OS_PART_POOL *part_group_head, *opool, *npool; USHORT part_group; USHORT i, j; char PoolName[8], *cp; sts = NU_Obtain_Semaphore(&MemSemCB, NU_SUSPEND); j = 0; part_group_head = 0; for (i = 0; i <= MaxPoolGroups; i++) { if (!PartGrpTable[i].grp_head || !PartGrpTable[i].name[0]) break; if (!strncmp(GroupName, PartGrpTable[i].name, RESOURCE_NAMELEN - 1)) { part_group_head = PartGrpTable[i].grp_head; opool = part_group_head; j++; while (opool->next) { opool = opool->next; j++; } break; } } /* * This error check logic has been modified from the original * faithful reconstruction by Space Falcon. In the original code * if MaxPoolGroups had been reached and the for loop above * never broke, the code would proceed to overwrite pool #0 * instead of catching the error. */ if (i > MaxPoolGroups) { release_sem_return_err: if (sts == NU_SUCCESS) NU_Release_Semaphore(&MemSemCB); return(OS_ERROR); } part_group = i; if (!part_group_head) { strncpy(PartGrpTable[part_group].name, GroupName, RESOURCE_NAMELEN); PartGrpTable[part_group].name[RESOURCE_NAMELEN-1] = 0; } if (os_AllocateMemory(OS_NOTASK, (T_VOID_STRUCT **) &npool, sizeof(T_OS_PART_POOL), OS_NO_SUSPEND, os_ext_pool_handle) != OS_OK) goto release_sem_return_err; sprintf(PoolName, "POOL%1d%1d", part_group + 1, j); Size &= ~3; npool->pool_mem = Addr; #if 0 /* * FreeCalypso: in our first-attempt gcc-built firmwares we needed to * bzero the PM_PCB before calling NU_Create_Partition_Pool() to * prevent the possibility of Nucleus error checker failing the call * because the signature word happens to be there already. The issue * arose because we were using "raw" memory sections that weren't * zeroed out on boot like standard .bss, but in TI's original * architecture everything is zeroed out on boot, so we don't need * this additional zeroing here. */ bzero(&npool->pcb, sizeof(NU_PARTITION_POOL)); #endif if (NU_Create_Partition_Pool(&npool->pcb, PoolName, npool->pool_mem, POOL_SIZE(Num, Size), Size + PT_CHKOVERHEAD + PPM_OVERHEAD, NU_FIFO) != NU_SUCCESS) goto release_sem_return_err; if (!part_group_head) PartGrpTable[part_group].grp_head = npool; else opool->next = npool; npool->size = Size; npool->next = 0; *GroupHandle = part_group; cp = (char *) npool->pool_mem; if (PoolBorder[part_group].Start >= cp) PoolBorder[part_group].Start = cp; cp += POOL_SIZE(Num, Size); if (PoolBorder[part_group].End < cp) PoolBorder[part_group].End = cp; os_InitPartitionCheck(npool); if (sts == NU_SUCCESS) NU_Release_Semaphore(&MemSemCB); return(OS_OK); } GLOBAL LONG os_CreatePartitionPool_fixed_pool_size(OS_HANDLE TaskHandle, char *GroupName, void *Addr, USHORT PoolSize, ULONG PartSize, OS_HANDLE *GroupHandle, ULONG *NumCreated) { USHORT num; num = PoolSize / (PartSize + PT_CHKOVERHEAD + PT_OVERHEAD); *NumCreated = num; return os_CreatePartitionPool(TaskHandle, GroupName, Addr, num, PartSize, GroupHandle); } GLOBAL LONG os_CreateMemoryPool(OS_HANDLE TaskHandle, char *Name, void *Addr, ULONG PoolSize, OS_HANDLE *PoolHandle) { STATUS sts; USHORT i; sts = NU_Obtain_Semaphore(&MemSemCB, NU_SUSPEND); for (i = 0; i < NumOfMemoryPools; i++) if (!strncmp(Name, MemPoolTable[i].name, RESOURCE_NAMELEN-1)) { *PoolHandle = i; if (sts == NU_SUCCESS) NU_Release_Semaphore(&MemSemCB); return(OS_OK); } if (i >= MaxMemoryPools) { release_sem_return_err: if (sts == NU_SUCCESS) NU_Release_Semaphore(&MemSemCB); return(OS_ERROR); } if (i) { if (os_AllocateMemory(OS_NOTASK, (T_VOID_STRUCT **) &MemPoolTable[i].pcb, sizeof(NU_MEMORY_POOL), OS_NO_SUSPEND, os_ext_pool_handle) != OS_OK) goto release_sem_return_err; #if 0 /* * FreeCalypso: in our first-attempt gcc-built firmwares we * needed to bzero the DM_PCB before calling * NU_Create_Memory_Pool() to prevent the possibility of * Nucleus error checker failing the call because the signature * word happens to be there already. The issue arose because * we were using "raw" memory sections that weren't zeroed out * on boot like standard .bss, but in TI's original architecture * everything is zeroed out on boot, so we don't need this * additional zeroing here. */ bzero(MemPoolTable[i].pcb, sizeof(NU_MEMORY_POOL)); #endif } if (NU_Create_Memory_Pool(MemPoolTable[i].pcb, Name, Addr, PoolSize, 4, NU_FIFO) != NU_SUCCESS) goto release_sem_return_err; strncpy(MemPoolTable[i].name, Name, RESOURCE_NAMELEN); MemPoolTable[i].name[RESOURCE_NAMELEN-1] = 0; *PoolHandle = i; NumOfMemoryPools++; if (sts == NU_SUCCESS) NU_Release_Semaphore(&MemSemCB); return(OS_OK); }