linux-metag.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Willy Tarreau <w@1wt.eu>
To: linux-kernel@vger.kernel.org, stable@vger.kernel.org, linux@roeck-us.net
Cc: James Hogan <james.hogan@imgtec.com>,
	linux-metag@vger.kernel.org, Jiri Slaby <jslaby@suse.cz>,
	Willy Tarreau <w@1wt.eu>
Subject: [PATCH 3.10 205/250] metag/usercopy: Fix src fixup in from user rapf loops
Date: Thu,  8 Jun 2017 00:59:51 +0200	[thread overview]
Message-ID: <1496876436-32402-206-git-send-email-w@1wt.eu> (raw)
In-Reply-To: <1496876436-32402-1-git-send-email-w@1wt.eu>

From: James Hogan <james.hogan@imgtec.com>

commit 2c0b1df88b987a12d95ea1d6beaf01894f3cc725 upstream.

The fixup code to rewind the source pointer in
__asm_copy_from_user_{32,64}bit_rapf_loop() always rewound the source by
a single unit (4 or 8 bytes), however this is insufficient if the fault
didn't occur on the first load in the loop, as the source pointer will
have been incremented but nothing will have been stored until all 4
register [pairs] are loaded.

Read the LSM_STEP field of TXSTATUS (which is already loaded into a
register), a bit like the copy_to_user versions, to determine how many
iterations of MGET[DL] have taken place, all of which need rewinding.

Fixes: 373cd784d0fc ("metag: Memory handling")
Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: linux-metag@vger.kernel.org
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Signed-off-by: Willy Tarreau <w@1wt.eu>
---
 arch/metag/lib/usercopy.c | 36 ++++++++++++++++++++++++++++--------
 1 file changed, 28 insertions(+), 8 deletions(-)

diff --git a/arch/metag/lib/usercopy.c b/arch/metag/lib/usercopy.c
index c3697a1..4adbab2 100644
--- a/arch/metag/lib/usercopy.c
+++ b/arch/metag/lib/usercopy.c
@@ -809,29 +809,49 @@ EXPORT_SYMBOL(__copy_user);
  *
  *	Rationale:
  *		A fault occurs while reading from user buffer, which is the
- *		source. Since the fault is at a single address, we only
- *		need to rewind by 8 bytes.
+ *		source.
  *		Since we don't write to kernel buffer until we read first,
  *		the kernel buffer is at the right state and needn't be
- *		corrected.
+ *		corrected, but the source must be rewound to the beginning of
+ *		the block, which is LSM_STEP*8 bytes.
+ *		LSM_STEP is bits 10:8 in TXSTATUS which is already read
+ *		and stored in D0Ar2
+ *
+ *		NOTE: If a fault occurs at the last operation in M{G,S}ETL
+ *			LSM_STEP will be 0. ie: we do 4 writes in our case, if
+ *			a fault happens at the 4th write, LSM_STEP will be 0
+ *			instead of 4. The code copes with that.
  */
 #define __asm_copy_from_user_64bit_rapf_loop(to, from, ret, n, id)	\
 	__asm_copy_user_64bit_rapf_loop(to, from, ret, n, id,		\
-		"SUB	%1, %1, #8\n")
+		"LSR	D0Ar2, D0Ar2, #5\n"				\
+		"ANDS	D0Ar2, D0Ar2, #0x38\n"				\
+		"ADDZ	D0Ar2, D0Ar2, #32\n"				\
+		"SUB	%1, %1, D0Ar2\n")
 
 /*	rewind 'from' pointer when a fault occurs
  *
  *	Rationale:
  *		A fault occurs while reading from user buffer, which is the
- *		source. Since the fault is at a single address, we only
- *		need to rewind by 4 bytes.
+ *		source.
  *		Since we don't write to kernel buffer until we read first,
  *		the kernel buffer is at the right state and needn't be
- *		corrected.
+ *		corrected, but the source must be rewound to the beginning of
+ *		the block, which is LSM_STEP*4 bytes.
+ *		LSM_STEP is bits 10:8 in TXSTATUS which is already read
+ *		and stored in D0Ar2
+ *
+ *		NOTE: If a fault occurs at the last operation in M{G,S}ETL
+ *			LSM_STEP will be 0. ie: we do 4 writes in our case, if
+ *			a fault happens at the 4th write, LSM_STEP will be 0
+ *			instead of 4. The code copes with that.
  */
 #define __asm_copy_from_user_32bit_rapf_loop(to, from, ret, n, id)	\
 	__asm_copy_user_32bit_rapf_loop(to, from, ret, n, id,		\
-		"SUB	%1, %1, #4\n")
+		"LSR	D0Ar2, D0Ar2, #6\n"				\
+		"ANDS	D0Ar2, D0Ar2, #0x1c\n"				\
+		"ADDZ	D0Ar2, D0Ar2, #16\n"				\
+		"SUB	%1, %1, D0Ar2\n")
 
 
 /* Copy from user to kernel, zeroing the bytes that were inaccessible in
-- 
2.8.0.rc2.1.gbe9624a


  parent reply	other threads:[~2017-06-07 22:59 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <1496876436-32402-1-git-send-email-w@1wt.eu>
2017-06-07 22:59 ` [PATCH 3.10 202/250] metag/usercopy: Fix alignment error checking Willy Tarreau
2017-06-07 22:59 ` [PATCH 3.10 203/250] metag/usercopy: Add early abort to copy_to_user Willy Tarreau
2017-06-07 22:59 ` [PATCH 3.10 204/250] metag/usercopy: Set flags before ADDZ Willy Tarreau
2017-06-07 22:59 ` Willy Tarreau [this message]
     [not found] ` <1496876436-32402-1-git-send-email-w-K+wRfnb2/UA@public.gmane.org>
2017-06-07 22:59   ` [PATCH 3.10 206/250] metag/usercopy: Add missing fixups Willy Tarreau

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1496876436-32402-206-git-send-email-w@1wt.eu \
    --to=w@1wt.eu \
    --cc=james.hogan@imgtec.com \
    --cc=jslaby@suse.cz \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-metag@vger.kernel.org \
    --cc=linux@roeck-us.net \
    --cc=stable@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).