linux-kernel - [PATCH] x86/fault: Print "SUPERVISOR" and "READ" when decoding #PF oops

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: <20181205163624.1842-1-sean.j.christopherson@intel.com>
Date:   Wed,  5 Dec 2018 08:36:24 -0800
From:   Sean Christopherson <sean.j.christopherson@...el.com>
To:     Dave Hansen <dave.hansen@...ux.intel.com>,
        Andy Lutomirski <luto@...nel.org>,
        Peter Zijlstra <peterz@...radead.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
        x86@...nel.org
Cc:     "H. Peter Anvin" <hpa@...or.com>, linux-kernel@...r.kernel.org,
        Linus Torvalds <torvalds@...ux-foundation.org>,
        Rik van Riel <riel@...riel.com>,
        Yu-cheng Yu <yu-cheng.yu@...el.com>,
        Ingo Molnar <mingo@...nel.org>
Subject: [PATCH] x86/fault: Print "SUPERVISOR" and "READ" when decoding #PF oops

...instead of manually handling the case where error_code=0, e.g. to
display "[SUPERVISOR] [READ]" instead of "normal kernel read fault".

This makes the zero case consistent with all other messages and also
provides additional information for other error code combinations,
e.g. error_code==1 will display "[PROT] [SUPERVISOR] [READ]" instead
of simply "[PROT]".

Print unique names for the negative cases as opposed to e.g. "[!USER]"
to avoid mixups due to users missing a single "!" character, and to be
more concise for the !INSTR && !WRITE case.

Print "SUPERVISOR" in favor of "KERNEL" to reduce the likelihood that
the message is misinterpreted as a generic kernel/software error and
to be consistent with the SDM's nomenclature.

An alternative to passing a negated error code to err_str_append() would
be to expand err_str_append() to take a second string for the negative
test, but that approach complicates handling the "[READ]" case, which
looks for !INSTR && !WRITE, e.g. it would require an extra call to
err_str_append() and logic in err_str_append() to allow null messages
for both the positive and negative tests.  Printing "[INSTR] [READ]"
wouldn't be the end of the world, but a little bit of trickery in the
kernel is a relatively small price to pay in exchange for the ability
to unequivocally know the access type by reading a single word.

Now that all components of the message use the [<code>] format,
explicitly state that it's the error *code* that's being printed and
group the err_str_append() calls by type so that the resulting print
messages are consistent, e.g. the deciphered codes will always be:

    [PROT] [USER|SUPERVISOR] [WRITE|INSTR|READ] [RSDV] [PK]

Cc: Andy Lutomirski <luto@...nel.org>
Cc: Borislav Petkov <bp@...en8.de>
Cc: Dave Hansen <dave.hansen@...ux.intel.com>
Cc: H. Peter Anvin <hpa@...or.com>
Cc: Linus Torvalds <torvalds@...ux-foundation.org>
Cc: Peter Zijlstra <peterz@...radead.org>
Cc: Rik van Riel <riel@...riel.com>
Cc: Thomas Gleixner <tglx@...utronix.de>
Cc: Yu-cheng Yu <yu-cheng.yu@...el.com>
Cc: linux-kernel@...r.kernel.org
Cc: Ingo Molnar <mingo@...nel.org>
Signed-off-by: Sean Christopherson <sean.j.christopherson@...el.com>
---
 arch/x86/mm/fault.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c
index 2ff25ad33233..0b4ce5d2b461 100644
--- a/arch/x86/mm/fault.c
+++ b/arch/x86/mm/fault.c
@@ -609,7 +609,7 @@ static void show_ldttss(const struct desc_ptr *gdt, const char *name, u16 index)
  */
 static void err_str_append(unsigned long error_code, char *buf, unsigned long mask, const char *txt)
 {
-	if (error_code & mask) {
+	if ((error_code & mask) == mask) {
 		if (buf[0])
 			strcat(buf, " ");
 		strcat(buf, txt);
@@ -655,13 +655,16 @@ show_fault_oops(struct pt_regs *regs, unsigned long error_code, unsigned long ad
 	 * zero delimiter must fit into err_txt[].
 	 */
 	err_str_append(error_code, err_txt, X86_PF_PROT,  "[PROT]" );
-	err_str_append(error_code, err_txt, X86_PF_WRITE, "[WRITE]");
 	err_str_append(error_code, err_txt, X86_PF_USER,  "[USER]" );
-	err_str_append(error_code, err_txt, X86_PF_RSVD,  "[RSVD]" );
+	err_str_append(~error_code, err_txt, X86_PF_USER, "[SUPERVISOR]");
+	err_str_append(error_code, err_txt, X86_PF_WRITE, "[WRITE]");
 	err_str_append(error_code, err_txt, X86_PF_INSTR, "[INSTR]");
+	err_str_append(~error_code, err_txt, X86_PF_WRITE | X86_PF_INSTR,
+							  "[READ]");
+	err_str_append(error_code, err_txt, X86_PF_RSVD,  "[RSVD]" );
 	err_str_append(error_code, err_txt, X86_PF_PK,    "[PK]"   );

-	pr_alert("#PF error: %s\n", error_code ? err_txt : "[normal kernel read fault]");
+	pr_alert("#PF error code: %s\n", err_txt);

 	if (!(error_code & X86_PF_USER) && user_mode(regs)) {
 		struct desc_ptr idt, gdt;
-- 
2.19.2