我正在使用带有Arduino Mega的Petit FatFS。我已经实现了所有必要的功能,但是pf_read()
功能有问题。我正在尝试读取一个568(512+56(字节长的文件。第一个扇区可以毫无问题地读取(而且是正确的(,但第二个部分扇区无法读取。
我试着调试pf_read()
,发现如下:对于第一个扇区,它用sector = 41104
、offset = 0
和count = 512
调用disk_readp()
,这似乎是正确的。然而,第二次它调用sector = -538935151
、offset = 512
和count = 56
(至少计数是正确的(。
奇怪的是,在pf_read()
完成之后,*br
就是97
,尽管函数实际上读取了512个字节。
为了防止disk_readp()
出现问题,下面是我的代码:
DRESULT disk_readp (
BYTE* buff, /* Pointer to the destination object */
DWORD sector, /* Sector number (LBA) */
UINT offset, /* Offset in the sector */
UINT count /* Byte count (bit15:destination) */
)
{
uint8_t r1;
uint8_t res_token;
uint16_t readAttempts;
DRESULT res = RES_ERROR;
CS_HIGH();
spi_transmit(0xFF);
CS_LOW();
spi_transmit(0xFF);
r1 = send_command(CMD17, sector);
if(r1 != 0xFF)
{
readAttempts = 0;
while(++readAttempts != SD_MAX_READ_ATTEMPTS)
{
if((res_token = spi_transmit(0xFF)) != 0xFF) break;
}
if(res_token == 0xFE)
{
// read 512 byte block
for(uint16_t i = 0; i < offset; i++) {
spi_transmit(0xFF); // discard data from (0) to (offset)
}
for(uint16_t i = offset; i < offset + count; i++)
{
*(buff++) = spi_transmit(0xFF); // safe data from (offset) to (offset + count)
}
for(uint16_t i = offset + count; i < 512; i++) {
spi_transmit(0xFF); // discard data from (offset + count) to (512)
}
// read and ignore 16-bit CRC
spi_transmit(0xFF);
spi_transmit(0xFF);
res = RES_OK;
}
}
CS_LOW();
spi_transmit(0xFF);
CS_HIGH();
spi_transmit(0xFF);
return res;
}
以及Petit FatFS文档的链接:http://elm-chan.org/fsw/ff/00index_p.html
谢谢
我终于发现了问题。由于某种原因,gcc
优化掉了pf_read()
的一个重要部分。
FRESULT pf_read (
void* buff, /* Pointer to the read buffer (NULL:Forward data to the stream)*/
UINT btr, /* Number of bytes to read */
UINT* br /* Pointer to number of bytes read */
)
{
DRESULT dr;
CLUST clst;
DWORD sect, remain;
UINT rcnt;
BYTE cs, *rbuff = buff;
FATFS *fs = FatFs;
*br = 0;
if (!fs) return FR_NOT_ENABLED; /* Check file system */
if (!(fs->flag & FA_OPENED)) return FR_NOT_OPENED; /* Check if opened */
remain = fs->fsize - fs->fptr;
if (btr > remain) btr = (UINT)remain; /* Truncate btr by remaining bytes */
while (btr) { /* Repeat until all data transferred */
if ((fs->fptr % 512) == 0) { /* On the sector boundary? */
cs = (BYTE)(fs->fptr / 512 & (fs->csize - 1)); /* Sector offset in the cluster */
if (!cs) { /* On the cluster boundary? */
if (fs->fptr == 0) { /* On the top of the file? */
clst = fs->org_clust;
} else {
clst = get_fat(fs->curr_clust);
}
if (clst <= 1) ABORT(FR_DISK_ERR);
fs->curr_clust = clst; /* Update current cluster */
}
sect = clust2sect(fs->curr_clust); /* Get current sector */
if (!sect) ABORT(FR_DISK_ERR);
fs->dsect = sect + cs;
}
rcnt = 512 - (UINT)fs->fptr % 512; /* Get partial sector data from sector buffer */
if (rcnt > btr) rcnt = btr;
debugInt(fs->dsect);
debugInt((UINT)fs->fptr % 512);
debugInt(rcnt);
writeString("n");
dr = disk_readp(rbuff, fs->dsect, (UINT)fs->fptr % 512, rcnt);
if (dr) ABORT(FR_DISK_ERR);
fs->fptr += rcnt; /* Advances file read pointer */
btr -= rcnt; *br += rcnt; /* Update read counter */
if (rbuff) rbuff += rcnt; /* Advances the data pointer if destination is memory */
}
return FR_OK;
}
*br = 0
行永远不会被执行,这完全打乱了进一步的计算。在我关闭了-O0
的优化后,一切都正常了。
因此,在将*br
传递给函数之前,我只确保将其设置为0。并不能真正解决问题,但至少它现在起作用了。
(GCC 11.1.0(
编辑:
没关系,我只是忘了指针是如何工作的。我这样叫pf_read()
:
unsigned int *br;
pf_read(path, size, br);