linux-kernel - [PATCH v9 06/18] arm64: mm: Always update TCR_EL1 from __cpu_set_tcr

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

Date:   Wed, 25 Mar 2020 23:24:08 -0400
From:   Pavel Tatashin <pasha.tatashin@...een.com>
To:     pasha.tatashin@...een.com, jmorris@...ei.org, sashal@...nel.org,
        ebiederm@...ssion.com, kexec@...ts.infradead.org,
        linux-kernel@...r.kernel.org, corbet@....net,
        catalin.marinas@....com, will@...nel.org,
        linux-arm-kernel@...ts.infradead.org, maz@...nel.org,
        james.morse@....com, vladimir.murzin@....com,
        matthias.bgg@...il.com, bhsharma@...hat.com, linux-mm@...ck.org,
        mark.rutland@....com, steve.capper@....com, rfontana@...hat.com,
        tglx@...utronix.de, selindag@...il.com
Subject: [PATCH v9 06/18] arm64: mm: Always update TCR_EL1 from __cpu_set_tcr_t0sz()

From: James Morse <james.morse@....com>

Because only the idmap sets a non-standard T0SZ, __cpu_set_tcr_t0sz()
can check for platforms that need to do this using
__cpu_uses_extended_idmap() before doing its work.

The idmap is only built with enough levels, (and T0SZ bits) to map
its single page.

To allow hibernate, and then kexec to idmap their single page copy
routines, __cpu_set_tcr_t0sz() needs to consider additional users,
who may need a different number of levels/T0SZ-bits to the idmap.
(i.e. VA_BITS may be enough for the idmap, but not hibernate/kexec)

Always read TCR_EL1, and check whether any work needs doing for
this request. __cpu_uses_extended_idmap() remains as it is used
by KVM, whose idmap is also part of the kernel image.

This mostly affects the cpuidle path, where we now get an extra
system register read .

CC: Lorenzo Pieralisi <Lorenzo.Pieralisi@....com>
CC: Sudeep Holla <sudeep.holla@....com>
Signed-off-by: James Morse <james.morse@....com>
Signed-off-by: Pavel Tatashin <pasha.tatashin@...een.com>
---
 arch/arm64/include/asm/mmu_context.h | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/arch/arm64/include/asm/mmu_context.h b/arch/arm64/include/asm/mmu_context.h
index 3827ff4040a3..09ecbfd0ad2e 100644
--- a/arch/arm64/include/asm/mmu_context.h
+++ b/arch/arm64/include/asm/mmu_context.h
@@ -79,16 +79,15 @@ static inline bool __cpu_uses_extended_idmap_level(void)
 }
 
 /*
- * Set TCR.T0SZ to its default value (based on VA_BITS)
+ * Ensure TCR.T0SZ is set to the provided value.
  */
 static inline void __cpu_set_tcr_t0sz(unsigned long t0sz)
 {
-	unsigned long tcr;
+	unsigned long tcr = read_sysreg(tcr_el1);
 
-	if (!__cpu_uses_extended_idmap())
+	if ((tcr & TCR_T0SZ_MASK) >> TCR_T0SZ_OFFSET == t0sz)
 		return;
 
-	tcr = read_sysreg(tcr_el1);
 	tcr &= ~TCR_T0SZ_MASK;
 	tcr |= t0sz << TCR_T0SZ_OFFSET;
 	write_sysreg(tcr, tcr_el1);
-- 
2.17.1