Hello Jaehoon,
I didn't think so..Our environment is support the CONFIG_SYS_MMC_MAX_BLK_COUNT.
This is defined in mmc.c right after the include definitions. The comment says that: Set block count limit because of 16 bit register limit on some hardware
So my use of this define is a bit of a hack too.
Did you know how get the timeout value "1000"?
If the timeout value "1000" is reasonable, i want to know what basis. Well, i don't think that my timeout value is reasonable.
I think timeout value was from original creator/maintainer of SD/MMC code.
When trying to read big data off of the SD card, I get the timeout too on my platform (ml507). My test consist of reading a 26Mb file from the SD card in PIO and DMA mode.
When the timeout is displayed in the serial console, the block count register is not yet zero but there is no error in the interrupt status register. The block count value shows the same value when I perform the same test.
For my use, CONFIG_SYS_MMC_MAX_BLK_COUNT = 0x1000 in my environment. I did not modify any of the timeout values.
Actually i want to remove the timeout value in that function. But then we should be prevent the infinite loop.
The infinite loop that you mention does not occur in my situation.
Anyway, thanks for your comment. I will check the your opinion.
HTH
(replying via gmane since i just joined the list)