Thread (7 messages) 7 messages, 5 authors, 2025-09-26

Re: [PATCH net v3] net: usb: Remove disruptive netif_wake_queue in rtl8150_set_multicast

From: Michal Pecio <hidden>
Date: 2025-09-24 20:42:47
Also in: linux-kernel-mentees, linux-usb, lkml

On Wed, 24 Sep 2025 19:50:55 +0200, Michal Pecio wrote:
Do you happen to remember what was the reason for padding all TX frames
to at least 60 bytes?

This was apparently added in version "v0.5.0 (2002/03/28)".

I'm yet to test the exact effect of this hack (will the HW really send
frames with trailing garbage?) and what happens if it's removed (maybe
nothing bad? or was there a HW bug?), but this part caught my attention
because I think nowadays some people could consider it "information
leak" ;) And it looks like a waste of bandwidth at least.
Sorry, stupid question, such frames are illegal.

That being said, I see that other drivers pad them with zeros or
other fixed pattern ('skb_padto(skb, ETH_ZLEN)' seems to be common)
rather than just DMA beyond the specified length.
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help