On Tue, Nov 27, 2018 at 4:19 AM Chris Packham judge.packham@gmail.com wrote:
On Mon, Nov 26, 2018 at 9:12 PM Simon Goldschmidt simon.k.r.goldschmidt@gmail.com wrote:
On Mon, Nov 26, 2018 at 9:00 AM Chris Packham judge.packham@gmail.com wrote:
ether_crc was added to the core net code in commit 53a5c424bf86 ("multicast tftp: RFC2090") so that other drivers could use it. However the only current user of it is tsec.c so move it there.
Signed-off-by: Chris Packham judge.packham@gmail.com
drivers/net/tsec.c | 25 +++++++++++++++++++++++++ include/net.h | 1 - net/eth_legacy.c | 24 ------------------------ 3 files changed, 25 insertions(+), 25 deletions(-)
diff --git a/drivers/net/tsec.c b/drivers/net/tsec.c index 03a46da2f8a1..9a4fab85e928 100644 --- a/drivers/net/tsec.c +++ b/drivers/net/tsec.c @@ -80,6 +80,31 @@ static void tsec_configure_serdes(struct tsec_private *priv)
#ifdef CONFIG_MCAST_TFTP
+/* the 'way' for ethernet-CRC-32. Spliced in from Linux lib/crc32.c
- and this is the ethernet-crc method needed for TSEC -- and perhaps
- some other adapter -- hash tables
- */
+#define CRCPOLY_LE 0xedb88320 +static u32 ether_crc(size_t len, unsigned char const *p)
I haven't checked, but can't we use lib/crc32.c for this? The polynomial is the same...
Yes more than likely. I erred on the side of not changing more than absolutely necessary. However given the fact that this code isn't currently run on any platform maybe that's too cautious.
A quick test shows me that a call to 'ether_crc(len, data)' can be replaced by 'bitrev32(crc32_no_comp(~0, data, len))'. Of course the code size increases as those files have tables (crc table: 1K, bitrev table: 256 bytes).
Simon
+{
int i;u32 crc;crc = ~0;while (len--) {crc ^= *p++;for (i = 0; i < 8; i++)crc = (crc >> 1) ^ ((crc & 1) ? CRCPOLY_LE : 0);}/* an reverse the bits, cuz of way they arrive -- last-first */crc = (crc >> 16) | (crc << 16);crc = (crc >> 8 & 0x00ff00ff) | (crc << 8 & 0xff00ff00);crc = (crc >> 4 & 0x0f0f0f0f) | (crc << 4 & 0xf0f0f0f0);crc = (crc >> 2 & 0x33333333) | (crc << 2 & 0xcccccccc);crc = (crc >> 1 & 0x55555555) | (crc << 1 & 0xaaaaaaaa);Does lib/bitrev.c do this job?
Regards, Simon
return crc;+}
/* CREDITS: linux gianfar driver, slightly adjusted... thanx. */
/* Set the appropriate hash bit for the given addr */ diff --git a/include/net.h b/include/net.h index 51c099dae2e5..359bfb5ef69f 100644 --- a/include/net.h +++ b/include/net.h @@ -289,7 +289,6 @@ const char *eth_get_name(void); /* get name of current device */
#ifdef CONFIG_MCAST_TFTP int eth_mcast_join(struct in_addr mcast_addr, int join); -u32 ether_crc(size_t len, unsigned char const *p); #endif
diff --git a/net/eth_legacy.c b/net/eth_legacy.c index 2a9caa3509b0..d2e16b8fa3da 100644 --- a/net/eth_legacy.c +++ b/net/eth_legacy.c @@ -310,30 +310,6 @@ int eth_mcast_join(struct in_addr mcast_ip, int join) return eth_current->mcast(eth_current, mcast_mac, join); }
-/* the 'way' for ethernet-CRC-32. Spliced in from Linux lib/crc32.c
- and this is the ethernet-crc method needed for TSEC -- and perhaps
- some other adapter -- hash tables
- */
-#define CRCPOLY_LE 0xedb88320 -u32 ether_crc(size_t len, unsigned char const *p) -{
int i;u32 crc;crc = ~0;while (len--) {crc ^= *p++;for (i = 0; i < 8; i++)crc = (crc >> 1) ^ ((crc & 1) ? CRCPOLY_LE : 0);}/* an reverse the bits, cuz of way they arrive -- last-first */crc = (crc >> 16) | (crc << 16);crc = (crc >> 8 & 0x00ff00ff) | (crc << 8 & 0xff00ff00);crc = (crc >> 4 & 0x0f0f0f0f) | (crc << 4 & 0xf0f0f0f0);crc = (crc >> 2 & 0x33333333) | (crc << 2 & 0xcccccccc);crc = (crc >> 1 & 0x55555555) | (crc << 1 & 0xaaaaaaaa);return crc;-}
#endif
-- 2.19.2
U-Boot mailing list U-Boot@lists.denx.de https://lists.denx.de/listinfo/u-boot