Hm, why not add seq_file support to dev_coredump? Neither git blame nor google sched any light on why seq_file wasn't picked over the custom read interface ...
Adding Johannes and Greg about this.
Main reason was that this is used for devcoredump which has its own similar but not quite seq_file compatible callback. If there is synergy to be had there that would be great because reinventing the wheel isn't fun.
Adding or changing it to seq_file is fine with me, I don't think we really need the devm_coredump() much these days since we have the vmalloc one.
(apologies for the footer and all - I'm on vacation and in a hurry)
johannes