Possible overflow when using BLOCK_BUFFER_SIZE >= 128

[FanDjango]

Possible overflow when using BLOCK_BUFFER_SIZE >= 128

When defining a BLOCK_BUFFER_SIZE >= 128 in configuration_adv.h in combination with using #define DISABLE_INACTIVE_EXTRUDER in configuration.h, there will be a compiler warning (with possible unwanted consequences) as follows:

  438 |       static uint8_t g_uc_extruder_last_move[EXTRUDERS];
      |                      ^~~~~~~~~~~~~~~~~~~~~~~
In file included from Marlin\src\module\../inc/../core/boards.h:24,
                 from Marlin\src\module\../inc/MarlinConfigPre.h:37,
                 from Marlin\src\module\../inc/MarlinConfig.h:28,
                 from Marlin\src\module\../MarlinCore.h:24,
                 from Marlin\src\module\planner.h:33,
                 from Marlin\src\module\planner.cpp:65:
Marlin\src\module\planner.cpp: In static member function 'static bool Planner::_populate_block(block_t*, bool, const abce_long_t&, const xyze_pos_t&, feedRate_t, uint8_t, const float&)':
Marlin\src\module\planner.cpp:2052:62: warning: unsigned conversion from 'int' to 'uint8_t' {aka 'unsigned char'} 
changes value from '256' to '0' [-Woverflow]
 2052 |             g_uc_extruder_last_move[N] = (BLOCK_BUFFER_SIZE) * 2; \
      |                                          ~~~~~~~~~~~~~~~~~~~~^~~

Changing the definition of g_uc_extruder_last_move from uint8_t to uint16_t in planner.cpp/planner.h would solve this and this won't cost very much memory (i.e. plus one byte per extruder).

[thinkyhead]

I'm not sure we want to advocate such a large buffer size. There's no special benefit to a very large buffer. I may just add a sanity check for buffer size larger than 64.

[FanDjango]

Sanity check would be sufficient, then the conditional type would not even be needed.

Need to modify the comment in configuration_adv imho.

[thinkyhead]

I don't want to make it a hard policy, per se. I do, however, want to suggest that for most users there is not a need to use such a large buffer.