Claude review: fbdev: Wrap user-invoked calls to fb_set_var() in helper

[Claude Code Review Bot <claude-review@example.com>]

Patch Review

Creates `fb_set_var_from_user()` in `fbmem.c` that bundles `fbcon_modechange_possible()` + `fb_set_var()` + `fbcon_update_vcs()` into a single call, then converts all callers.

The chrdev ioctl path (`fb_chrdev.c`) was already doing all three steps — this is a direct refactor with no behavior change.

The sysfs `activate()` path (`fbsysfs.c`) was missing the `fbcon_modechange_possible()` check. The new helper adds it, which is a correctness improvement — mode changes through sysfs will now be properly validated against fbcon constraints.

The `ps3fb` driver was also missing the `fbcon_modechange_possible()` check, which is similarly a correctness improvement.

**Semantic change in `sh_mobile_lcdcfb`:** The original code hardcoded `fbcon_update_vcs(info, true)` (update all VCs). The new helper passes `var->activate & FB_ACTIVATE_ALL`. Since `sh_mobile_fb_reconfig()` sets `var.activate = FB_ACTIVATE_NOW` (which is 0), `FB_ACTIVATE_NOW & FB_ACTIVATE_ALL` evaluates to `false`, meaning only the foreground VC gets updated instead of all VCs. This is a functional change — whether it's intentional or a regression depends on the driver's requirements. It's worth documenting in the commit message if intentional, or fixing if not.

The removal of `EXPORT_SYMBOL` for `fbcon_update_vcs` and `EXPORT_SYMBOL_GPL` for `fbcon_modechange_possible` is correct since patches 1-3 eliminate all out-of-module callers before patch 4 fully internalizes them.

The new `fb_set_var_from_user` is `EXPORT_SYMBOL` which is appropriate since `ps3fb` and `sh_mobile_lcdcfb` are separate drivers that need it.

---
Generated by Claude Code Patch Reviewer