g2gps

Adding a larger stream buffer changes the behaviour for fatfs. Setting `stdio_buffer_size = 4096`: ``` open test_file with mode w set stdio buffer to: 0x80248290 size: 4096 file position 0...

I still have an issue when switching between littlefs and fatfs. However, I'm currently unable to come up with a reproducible test case.

> @g2gps only kernel mode has this problem? Yes, it seems to be just kernel mode. It works with ` rv-virt:libcxx` using `c++2b` and `gnu++20`.

@xiaoxiang781216 Apologies I was out of the NuttX space. I've update our local mirror to include #13721 and can confirm that it resolves this issue.

> @g2gps any idea why for 64KB block the write performance with copy it worst? Is it because maybe cache misses? It's because the amount of data copied is larger...

> how about port sdbench and sdstress back to nuttx/apps? I don't mind pushing what I have back up, I'm just not sure on licencing requirements. Do we need to...

@Donny9 Final results of `fstest -n 10 -m /mnt -o 512 -s 8192`: ``` Final memory usage: VARIABLE BEFORE AFTER DELTA ======== ======== ======== ======== arena 81ffc 81ffc 0 ordblks...

> BTW, please add an entry in our https://nuttx.apache.org/docs/latest/components/drivers/block/index.html# Doc to say this feature exists (otherwise it will be a "hidden feature") Sorry, I just realized that I missed this....

@FASTSHIFT Thanks for trying this out. A couple of things - Would you mind enabling `CONFIG_BLOCK_CACHE_DEBUG` and rerunning that example, so we can see the cache is being used. -...

I was hoping to revisit this and look at getting this integrated. We've been using this locally on a few of our targets since this PR was opened. One thing...