On Sun, Jun 21, 2020 at 08:27:45PM +0800, Li Guifu via Linux-erofs wrote: ...
> nocompression: > - ret = write_uncompressed_block(ctx, &len, dst); > + /* reset clusterofs to 0 if permitted */ > + if (!erofs_sb_has_lz4_0padding() && > + ctx->head >= ctx->clusterofs) { > + ctx->head -= ctx->clusterofs; > + len += ctx->clusterofs; > + limit += ctx->clusterofs; > + ctx->clusterofs = 0; TL;DR: it seems still buggy here. Pls rethink carefully and send a usable patch then... Thanks, Gao Xiang