shell/vfs: Handle print failure in genfile cmd gracefully

[Teufelchen1]

Contribution description

Hey 🐡

snprintf turned out to be difficult to get right.

Testing procedure

Try make -C tests/sys/vfs_default/ all-asan term and play with `genfile

Issues/PRs references

[riot-ci]

Murdock results

✔️ PASSED

8bd1de8 shell/vfs: Handle print failure in genfile cmd gracefully

Success	Failures	Total	Runtime
10065	0	10066	18m:59s

Artifacts

• Documentation preview

[mguetschow]

Thanks! This means we just silently ignore the truncation if it happens?

[kfessel]

the return value of this is 5 %03u is constant width unless it failed -> bs needs to be 6 to have blocks big enough ( fit this print and a /n )

this would be more readable (to me) if instead of pointer arithmetics, adresses of array positions would be used (in stead of *buf have unsigned pos)

*buf + size

vs

&blocks[pos]

[Teufelchen1]

Hm not entirely sure but while thinking of it, I found a neat little optimization. I moved the memset into the if-clause, please have a look.

[Teufelchen1]

This means we just silently ignore the truncation if it happens

Yes, I am not sure what would be best, so I thought: Not writing to a fs is better than writing truncated garbage to the fs.

[kfessel]

handles snprintf return value better than before avoiding mishaps

[kfessel]

expected results:

(any, 3, 8) will write "|0\n"
(any, 5, 8) will write "|008\n"
(any, 6, 8) will write "|008|\n"
(any, 9, 8) will write "|008|\0\0\0\n"

(any, 3, 8000) will write "|8\n"
(any, 5, 8000) will write "|800\n"
(any, 6, 8000) will write "|8000\n"
(any, 9, 8000) will write "|8000|\0\0\n"