Some comments after reading the I2C slave core guide

e8f4061e · Javier Serrano · c5561f06 · e8f4061e
Commit e8f4061e authored Nov 04, 2013 by Javier Serrano
Hide whitespace changes
Inline Side-by-side

Showing with 18 additions and 0 deletions

javier.txt design-review/javier.txt +18 -0

No files found.
--- a/design-review/javier.txt
+++ b/design-review/javier.txt
@@ -16,6 +16,24 @@ cycles? A maximum g_pwidth should also be imposed to preserve the
 transformer. "The behavior of the outputs are different" -> "The
 behavior of the outputs is different"

+Some comments while reading the "I2C Slave Core" document
+=========================================================
+
+I find it quite misleading that sometimes a "read" is from the slave's
+point of view, sometime from the master's. There should be only one
+convention, which as far as I'm concerned should be the master's point
+of view. A write is always a write from the master. Same for
+reads. This convention (which is basically the one used in all
+literature about VME, PCI, etc) should be stated just once at the
+beginning of the document.
+
+Page 5: we're asking the user of this core to set ack_n_i before SCL
+goes high. I think this is not a reasonable requirement because the
+user of this core has no control over SCL. In fact, shielding the user
+from any need to look at SCL should be one of the goals of this
+core. You can make a more reasonable request to the user by asking
+that ack_n_i be set one clk_i tick after done_p_o goes high, for
+example. 

 Todo
 ====