mac80211: add utility function to get header length
PATCH 3/7 mac80211: add utility function to get header length
Take a __le16 directly rather than a host-endian value.
--- include/net/mac80211.h | 6 ++++++ net/mac80211/util.c | 32 ++++++++++++++++++++++++++++++++ 2 files changed, 38 insertions(+), 0 deletions(-)
include/net/mac80211.h +6 -0
@@ -1548,6 +1548,12 @@ int ieee80211_get_hdrlen_from_skb(const struct sk_buff *skb);int ieee80211_get_hdrlen (u16 fc);/** +* ieee80211_hdrlen - get header length in bytes from frame control +* @fc: frame control field in little-endian format +*/ +unsigned int ieee80211_hdrlen (__le16 fc); + +/** * ieee80211_get_tkip_key - get a TKIP rc4 for skb * * This function computes a TKIP rc4 key for an skb. It computes
net/mac80211/util.c +32 -0
@@ -133,6 +133,38 @@ int ieee80211_get_hdrlen(u16 fc) }EXPORT_SYMBOL (ieee80211_get_hdrlen); +unsigned int ieee80211_hdrlen (__le16 fc) +{ +unsigned int hdrlen =24 ; + +if (ieee80211_is_data (fc)) { +if (ieee80211_has_a4 (fc)) + hdrlen =30 ; +if (ieee80211_data_has_qos (fc)) + hdrlen +=IEEE80211_QOS_CTL_LEN ; + goto out; + } + +if (ieee80211_is_ctl (fc)) { +/* +* ACK and CTS are 10 bytes, all others 16. To see how +* to get this condition consider +* subtype mask: 0b0000000011110000 (0x00F0) +* ACK subtype: 0b0000000011010000 (0x00D0) +* CTS subtype: 0b0000000011000000 (0x00C0) +* bits that matter: ^^^ (0x00E0) +* value of those: 0b0000000011000000 (0x00C0) +*/ +if ((fc &cpu_to_le16 (0x00E0 )) ==cpu_to_le16 (0x00C0 )) + hdrlen =10 ; +else + hdrlen =16 ; + } +out: +return hdrlen; +} +EXPORT_SYMBOL (ieee80211_hdrlen); +int ieee80211_get_hdrlen_from_skb (const struct sk_buff *skb) {const struct ieee80211_hdr *hdr = (const struct ieee80211_hdr *) skb->data;
Signature · 1 line
1.5.6.rc1.257.gba91d
Johannes Berg
maintainerinclude/net/mac80211.h int ieee80211_get_hdrlen_from_skb(const struct sk_buff *skb);
Quoted 9 lines
/**
+ *ieee80211_hdrlen- get header length in bytes from frame control
+ * @fc: frame control field in little-endian format
+ */include/net/mac80211.h +3
-0+unsigned int ieee80211_hdrlen (__le16 fc); + +EXPORT_SYMBOL (ieee80211_hdrlen);
Do we really need to export that?
johannes
Quoted 11 lines · Johannes Berg
include/net/mac80211.h int ieee80211_get_hdrlen_from_skb(const struct sk_buff *skb);
Quoted 9 lines
/**
+ *ieee80211_hdrlen- get header length in bytes from frame control
+ * @fc: frame control field in little-endian format
+ */include/net/mac80211.h +3
-0+unsigned int ieee80211_hdrlen (__le16 fc); + +EXPORT_SYMBOL (ieee80211_hdrlen);Do we really need to export that?
so now we have hdrlen(u16) and hdrlen(__le16) it will be fun.
So I guess will we converting idioms
u16 fc = le16_to_cpu(hdr->frame_control); int hdr_len = ieee80211_get_hdrlen(fc);
to
int hdr_len = ieee80211_hdrlen(hdr->frame_control)
This is how it used in driver code so it make sense to export this function and remove ieee80211_get_hdrlen(fc)
Since all fc operations are bitwise 'and' and 'or' u16 rx->fc can be dropped in future as well
Tomas
Quoted 16 lines · Tomas Winkler
Quoted 11 lines · Johannes Berg
include/net/mac80211.h int ieee80211_get_hdrlen_from_skb(const struct sk_buff *skb);
Quoted 9 lines
/**
+ *ieee80211_hdrlen- get header length in bytes from frame control
+ * @fc: frame control field in little-endian format
+ */include/net/mac80211.h +3
-0+unsigned int ieee80211_hdrlen (__le16 fc); + +EXPORT_SYMBOL (ieee80211_hdrlen);Do we really need to export that?
so now we have hdrlen(u16) and
hdrlen(__le16)it will be fun.
Quoted 8 lines
So I guess will we converting idioms
u16 fc =le16_to_cpu(hdr->frame_control); inthdr_len=ieee80211_get_hdrlen(fc);
to
inthdr_len=ieee80211_hdrlen(hdr->frame_control)This is how it used in driver code so it make sense to export this function and remove
ieee80211_get_hdrlen(fc)
Yes, that was my thinking, I just did it this way to avoid the flag day change, I'll trickle the changes in and then remove _get_hdrlen.
Quoted 3 lines
Since all fc operations are bitwise 'and' and 'or' u16
rx->fccan be dropped in future as well
I was going to convert it to a __le16, but if it is just a copy of the ->frame_control in the header, I'll look at removing it instead.
Cheers,
Harvey
Johannes Berg
maintainerQuoted 7 lines
Quoted 7 lines
So I guess will we converting idioms
u16 fc =le16_to_cpu(hdr->frame_control); inthdr_len=ieee80211_get_hdrlen(fc);
to
inthdr_len=ieee80211_hdrlen(hdr->frame_control)
Quoted 5 lines
Quoted 2 lines
This is how it used in driver code so it make sense to export this function and remove
ieee80211_get_hdrlen(fc)Yes, that was my thinking, I just did it this way to avoid the flag day change, I'll trickle the changes in and then remove
_get_hdrlen.
Sounds good to me.
Quoted 5 lines
Quoted 2 lines
Since all fc operations are bitwise 'and' and 'or' u16
rx->fccan be dropped in future as wellI was going to convert it to a
__le16, but if it is just a copy of the ->frame_control in the header, I'll look at removing it instead.
Yeah, I think rx->fc was meant to be a cpu-byteorder copy of hdr->frame_control to avoid repeated byteswapping. If we do all operations on the constants instead as you're doing with this series, we ought to be able to remove it. Bonus point: that gets rid of the possible "rx->fc is out of sync" issue we had to fix once already.
johannes
Quoted 6 lines · Johannes Berg
Quoted 6 lines
Quoted 6 lines
So I guess will we converting idioms
u16 fc =le16_to_cpu(hdr->frame_control); inthdr_len=ieee80211_get_hdrlen(fc);
to
inthdr_len=ieee80211_hdrlen(hdr->frame_control)
Quoted 8 lines
Quoted 5 lines
Quoted 2 lines
This is how it used in driver code so it make sense to export this function and remove
ieee80211_get_hdrlen(fc)Yes, that was my thinking, I just did it this way to avoid the flag day change, I'll trickle the changes in and then remove
_get_hdrlen.Sounds good to me.
OK, will keep going on this then.
Quoted 11 lines
Quoted 5 lines
Quoted 2 lines
Since all fc operations are bitwise 'and' and 'or' u16
rx->fccan be dropped in future as wellI was going to convert it to a
__le16, but if it is just a copy of the ->frame_control in the header, I'll look at removing it instead.Yeah, I think
rx->fcwas meant to be a cpu-byteorder copy ofhdr->frame_controlto avoid repeated byteswapping. If we do all operations on the constants instead as you're doing with this series, we ought to be able to remove it. Bonus point: that gets rid of the possible "rx->fcis out of sync" issue we had to fix once already.
That and be arches eliminate a bunch of byteswaps.
Cheers,
Harvey