update stream relative, fix mtd&blk stream readback before block update. decrese int use in stream to handle FS_LARGEFILE

[jasonbu]

Summary

int is not reliable when we use same code for cross platform.
replace with off_t,size_t,ssize_t should prefer.

offset int -> offset, len int -> size_t, ret int -> ssize_t

Also we should readback sector/block before operate to new sector/block,
for user only update partial case.

Impact

stream will able to support FS_LARGEFILE better cover >4GB devices.
if user use blk/mtd stream to update partial, the old data will be kept back.

Testing

CI-test, local arm-v8m prject, qemu-armv8a, ubutun sim:nsh

[nuttxpr]

[Experimental Bot, please feedback here]

Yes, this PR appears to meet the NuttX requirements, although the provided information is minimal. It would be stronger with more detail. Here's a breakdown:

Strengths:

• Clear Summary of "Why": Explains the motivation for the change (cross-platform compatibility and partial update handling).
• Functional Change Description: Identifies the affected code (int to off_t/size_t/ssize_t) and the readback mechanism.
• Impact on User/Features: Briefly mentions improved large file support and data preservation.
• Testing Performed: Lists tested platforms (CI, ARMv8-M, qemu, Ubuntu sim).

Weaknesses & Missing Information:

• No Issue References: Missing links to related NuttX issues. Even if there isn't a directly related issue, creating one beforehand is good practice.
• Impact Details Lacking: The "Impact" section needs significant expansion. While "YES" is stated for user and feature impact, the descriptions are very brief. More details are needed for all impact categories (build, hardware, documentation, security, compatibility). Be explicit about any potential breaking changes. For example, does the change to size_t and off_t affect any existing drivers or applications? If documentation is required, state whether it's included in the PR.
• Vague Testing Logs: Providing actual log snippets is crucial. "CI-test" is insufficient – link to the CI run. Include specific commands used and relevant output demonstrating the issue before the change and the corrected behavior afterward. Quantify the improvements if possible (e.g., "Successfully wrote a 5GB file," "Partial update preserved data from previous sectors X to Y").
• Build Host Details Missing: Specify your development environment in the "Testing" section as requested by the template (OS, CPU, compiler versions).

Recommendation:

While the core elements are present, this PR description needs more detail to ensure thorough review and avoid potential problems. Expand the "Impact" and "Testing" sections considerably, providing concrete examples and data. Add issue references. A more detailed PR will increase confidence in the changes and streamline the review process.

[xiaoxiang781216]

has the test fail:
https://github.com/apache/nuttx/actions/runs/11830852498/job/32965096269?pr=14778