linux-kernel - 2.6.29-rc libata sff 32bit PIO regression

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-ID: <Pine.LNX.4.64.0901211322010.15945@blonde.anvils>
Date:	Wed, 21 Jan 2009 13:34:39 +0000 (GMT)
From:	Hugh Dickins <hugh@...itas.com>
To:	Alan Cox <alan@...rguk.ukuu.org.uk>
cc:	Jeff Garzik <jgarzik@...hat.com>,
	"Rafael J. Wysocki" <rjw@...k.pl>, linux-ide@...r.kernel.org,
	linux-kernel@...r.kernel.org
Subject: 2.6.29-rc libata sff 32bit PIO regression

I've a Dell Precision 670 here (four-year-old EM64T Xeon with ata_piix)
which doesn't like your commit 871af1210f13966ab911ed2166e4ab2ce775b99d
libata: Add 32bit PIO support.  Full dmesg (and .config) attached, but
here's an extract showing the start of the error messages on ata2:

ata1: PATA max UDMA/100 cmd 0x1f0 ctl 0x3f6 bmdma 0xffa0 irq 14
ata2: PATA max UDMA/100 cmd 0x170 ctl 0x376 bmdma 0xffa8 irq 15
ata1: port disabled. ignoring.
ata2.00: ATAPI: GCR-8483B, 1.07, max UDMA/33
ata2.00: configured for UDMA/33
scsi 1:0:0:0: CD-ROM            HL-DT-ST CD-ROM GCR-8483B 1.07 PQ: 0 ANSI: 5
ata2.00: qc timeout (cmd 0xa0)
ata2.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x6 frozen
ata2.00: cmd a0/00:00:00:00:00/00:00:00:00:00/a0 tag 0
         cdb 00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00
         res 51/20:03:00:00:00/00:00:00:00:00/a0 Emask 0x5 (timeout)
ata2.00: status: { DRDY ERR }
ata2: soft resetting link
ata2.00: configured for UDMA/33
ata2: EH complete

linux-next behaved the same, so I played around, and the patch below seems
to fix it fine (though I've not yet tried it on other, working machines);
but I can well imagine you may prefer to do it differently.  Or was I just
wrong to move that box's CD handling from ide to libata a few months ago?

[PATCH] libata sff: 32bit PIO use 16bit on slop

871af1210f13966ab911ed2166e4ab2ce775b99d libata: Add 32bit PIO support
causes errors on a four-year-old ata_piix Dell Precision 670.  Using
16bit PIO instead of 32bit PIO on the odd 1, 2 or 3 chars fixes that.

Signed-off-by: Hugh Dickins <hugh@...itas.com>
---

 drivers/ata/libata-sff.c |   18 +++++++++++-------
 1 file changed, 11 insertions(+), 7 deletions(-)

--- 2.6.29-rc2/drivers/ata/libata-sff.c	2009-01-17 10:40:12.000000000 +0000
+++ linux/drivers/ata/libata-sff.c	2009-01-21 12:32:01.000000000 +0000
@@ -774,17 +774,21 @@ unsigned int ata_sff_data_xfer32(struct 
 		iowrite32_rep(data_addr, buf, words);
 
 	if (unlikely(slop)) {
-		__le32 pad;
+		unsigned char slop_buf[4];
+		unsigned char *trailing_buf = buf + buflen - slop;
+
+		words = (slop + 1) >> 1;
 		if (rw == READ) {
-			pad = cpu_to_le32(ioread32(ap->ioaddr.data_addr));
-			memcpy(buf + buflen - slop, &pad, slop);
+			ioread16_rep(data_addr, slop_buf, words);
+			memcpy(trailing_buf, slop_buf, slop);
 		} else {
-			memcpy(&pad, buf + buflen - slop, slop);
-			iowrite32(le32_to_cpu(pad), ap->ioaddr.data_addr);
+			memcpy(slop_buf, trailing_buf, slop);
+			slop_buf[slop] = 0;
+			iowrite16_rep(data_addr, slop_buf, words);
 		}
-		words++;
 	}
-	return words << 2;
+
+	return buflen + (slop & 1);
 }
 EXPORT_SYMBOL_GPL(ata_sff_data_xfer32);
 
View attachment "dmesg.empty" of type "TEXT/PLAIN" (23681 bytes)

View attachment "config.empty" of type "TEXT/PLAIN" (36611 bytes)