> Are client drivers setting this today, will there be regression if not set? I don't see any dw-axi-dmac client driver setting this src_maxburt/dst_maxburst explicitly. This means that all the while, those client drivers have been using the default value of DWAXIDMAC_BURST_TRANS_LEN_4. In the patch, to maintain this backward compatibility, ternary operator is used, for example: > > + dst_burst_trans_len = chan->config.dst_maxburst ? > > + __ffs(chan->config.dst_maxburst) - 1 : > > + DWAXIDMAC_BURST_TRANS_LEN_4; so that if src_maxburst/dst_maxburst is not set, default value of DWAXIDMAC_BURST_TRANS_LEN_4 applies. > this is wrong, memcpy should never use slave config value. These values are > for peripheral and not meant for mem-mem transfers Agreed, thanks for pointing that out. It shall be removed in patch v3.
