linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH 0/3] RISC-V: Fix cpumask rework falloout
@ 2022-01-31 13:26 Geert Uytterhoeven
  2022-01-31 13:26 ` [PATCH 1/3] RISC-V: Fix hartid mask handling for hartid 31 and up Geert Uytterhoeven
                   ` (3 more replies)
  0 siblings, 4 replies; 7+ messages in thread
From: Geert Uytterhoeven @ 2022-01-31 13:26 UTC (permalink / raw)
  To: Paul Walmsley, Palmer Dabbelt, Albert Ou, Anup Patel, Atish Patra
  Cc: Jessica Clarke, Emil Renner Berthing, linux-riscv, linux-kernel,
	Geert Uytterhoeven

	Hi all,

This patch series intends to fix the fall-out from commit
26fb751ca37846c9 ("RISC-V: Do not use cpumask data structure for hartid
bitmap"), cfr. the discussion in thread[1].

Thanks for your comments!

[1] "[PATCH v3 6/6] RISC-V: Do not use cpumask data structure for hartid
     bitmap"
    https://lore.kernel.org/r/20220120090918.2646626-7-atishp@rivosinc.com/

Geert Uytterhoeven (3):
  RISC-V: Fix hartid mask handling for hartid 31 and up
  RISC-V: Fix handling of empty cpu masks
  RISC-V: Fix IPI/RFENCE hmask on non-monotonic hartid ordering

 arch/riscv/kernel/sbi.c | 72 +++++++++++++++++++++++++++--------------
 1 file changed, 47 insertions(+), 25 deletions(-)

-- 
2.25.1

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds

^ permalink raw reply	[flat|nested] 7+ messages in thread

* [PATCH 1/3] RISC-V: Fix hartid mask handling for hartid 31 and up
  2022-01-31 13:26 [PATCH 0/3] RISC-V: Fix cpumask rework falloout Geert Uytterhoeven
@ 2022-01-31 13:26 ` Geert Uytterhoeven
  2022-01-31 13:26 ` [PATCH 2/3] RISC-V: Fix handling of empty cpu masks Geert Uytterhoeven
                   ` (2 subsequent siblings)
  3 siblings, 0 replies; 7+ messages in thread
From: Geert Uytterhoeven @ 2022-01-31 13:26 UTC (permalink / raw)
  To: Paul Walmsley, Palmer Dabbelt, Albert Ou, Anup Patel, Atish Patra
  Cc: Jessica Clarke, Emil Renner Berthing, linux-riscv, linux-kernel,
	Geert Uytterhoeven

Jessica reports that using "1 << hartid" causes undefined behavior for
hartid 31 and up.

Fix this by using the BIT() helper instead of an explicit shift.

Reported-by: Jessica Clarke <jrtc27@jrtc27.com>
Fixes: 26fb751ca37846c9 ("RISC-V: Do not use cpumask data structure for hartid bitmap")
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
---
 arch/riscv/kernel/sbi.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/arch/riscv/kernel/sbi.c b/arch/riscv/kernel/sbi.c
index f72527fcb34709a6..f93bc75f2bc43c8e 100644
--- a/arch/riscv/kernel/sbi.c
+++ b/arch/riscv/kernel/sbi.c
@@ -5,6 +5,7 @@
  * Copyright (c) 2020 Western Digital Corporation or its affiliates.
  */
 
+#include <linux/bits.h>
 #include <linux/init.h>
 #include <linux/pm.h>
 #include <linux/reboot.h>
@@ -85,7 +86,7 @@ static unsigned long __sbi_v01_cpumask_to_hartmask(const struct cpumask *cpu_mas
 			pr_warn("Unable to send any request to hartid > BITS_PER_LONG for SBI v0.1\n");
 			break;
 		}
-		hmask |= 1 << hartid;
+		hmask |= BIT(hartid);
 	}
 
 	return hmask;
@@ -268,7 +269,7 @@ static int __sbi_send_ipi_v02(const struct cpumask *cpu_mask)
 		}
 		if (!hmask)
 			hbase = hartid;
-		hmask |= 1UL << (hartid - hbase);
+		hmask |= BIT(hartid - hbase);
 	}
 
 	if (hmask) {
@@ -362,7 +363,7 @@ static int __sbi_rfence_v02(int fid, const struct cpumask *cpu_mask,
 		}
 		if (!hmask)
 			hbase = hartid;
-		hmask |= 1UL << (hartid - hbase);
+		hmask |= BIT(hartid - hbase);
 	}
 
 	if (hmask) {
-- 
2.25.1


^ permalink raw reply related	[flat|nested] 7+ messages in thread

* [PATCH 2/3] RISC-V: Fix handling of empty cpu masks
  2022-01-31 13:26 [PATCH 0/3] RISC-V: Fix cpumask rework falloout Geert Uytterhoeven
  2022-01-31 13:26 ` [PATCH 1/3] RISC-V: Fix hartid mask handling for hartid 31 and up Geert Uytterhoeven
@ 2022-01-31 13:26 ` Geert Uytterhoeven
  2022-01-31 21:16   ` Atish Patra
  2022-01-31 13:26 ` [PATCH 3/3] RISC-V: Fix IPI/RFENCE hmask on non-monotonic hartid ordering Geert Uytterhoeven
  2022-01-31 17:52 ` [PATCH 0/3] RISC-V: Fix cpumask rework falloout Emil Renner Berthing
  3 siblings, 1 reply; 7+ messages in thread
From: Geert Uytterhoeven @ 2022-01-31 13:26 UTC (permalink / raw)
  To: Paul Walmsley, Palmer Dabbelt, Albert Ou, Anup Patel, Atish Patra
  Cc: Jessica Clarke, Emil Renner Berthing, linux-riscv, linux-kernel,
	Geert Uytterhoeven

The cpumask rework slightly changed the behavior of the code.  Fix this
by treating an empty cpumask as meaning all online CPUs.

Extracted from a patch by Atish Patra <atishp@rivosinc.com>.

Reported-by: Jessica Clarke <jrtc27@jrtc27.com>
Fixes: 26fb751ca37846c9 ("RISC-V: Do not use cpumask data structure for hartid bitmap")
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
---
 arch/riscv/kernel/sbi.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/arch/riscv/kernel/sbi.c b/arch/riscv/kernel/sbi.c
index f93bc75f2bc43c8e..22444cfcd56cc646 100644
--- a/arch/riscv/kernel/sbi.c
+++ b/arch/riscv/kernel/sbi.c
@@ -161,7 +161,7 @@ static int __sbi_send_ipi_v01(const struct cpumask *cpu_mask)
 {
 	unsigned long hart_mask;
 
-	if (!cpu_mask)
+	if (!cpu_mask || cpumask_empty(cpu_mask))
 		cpu_mask = cpu_online_mask;
 	hart_mask = __sbi_v01_cpumask_to_hartmask(cpu_mask);
 
@@ -177,7 +177,7 @@ static int __sbi_rfence_v01(int fid, const struct cpumask *cpu_mask,
 	int result = 0;
 	unsigned long hart_mask;
 
-	if (!cpu_mask)
+	if (!cpu_mask || cpumask_empty(cpu_mask))
 		cpu_mask = cpu_online_mask;
 	hart_mask = __sbi_v01_cpumask_to_hartmask(cpu_mask);
 
@@ -254,7 +254,7 @@ static int __sbi_send_ipi_v02(const struct cpumask *cpu_mask)
 	struct sbiret ret = {0};
 	int result;
 
-	if (!cpu_mask)
+	if (!cpu_mask || cpumask_empty(cpu_mask))
 		cpu_mask = cpu_online_mask;
 
 	for_each_cpu(cpuid, cpu_mask) {
@@ -348,7 +348,7 @@ static int __sbi_rfence_v02(int fid, const struct cpumask *cpu_mask,
 	unsigned long hartid, cpuid, hmask = 0, hbase = 0;
 	int result;
 
-	if (!cpu_mask)
+	if (!cpu_mask || cpumask_empty(cpu_mask))
 		cpu_mask = cpu_online_mask;
 
 	for_each_cpu(cpuid, cpu_mask) {
-- 
2.25.1


^ permalink raw reply related	[flat|nested] 7+ messages in thread

* [PATCH 3/3] RISC-V: Fix IPI/RFENCE hmask on non-monotonic hartid ordering
  2022-01-31 13:26 [PATCH 0/3] RISC-V: Fix cpumask rework falloout Geert Uytterhoeven
  2022-01-31 13:26 ` [PATCH 1/3] RISC-V: Fix hartid mask handling for hartid 31 and up Geert Uytterhoeven
  2022-01-31 13:26 ` [PATCH 2/3] RISC-V: Fix handling of empty cpu masks Geert Uytterhoeven
@ 2022-01-31 13:26 ` Geert Uytterhoeven
  2022-01-31 21:18   ` Atish Patra
  2022-01-31 17:52 ` [PATCH 0/3] RISC-V: Fix cpumask rework falloout Emil Renner Berthing
  3 siblings, 1 reply; 7+ messages in thread
From: Geert Uytterhoeven @ 2022-01-31 13:26 UTC (permalink / raw)
  To: Paul Walmsley, Palmer Dabbelt, Albert Ou, Anup Patel, Atish Patra
  Cc: Jessica Clarke, Emil Renner Berthing, linux-riscv, linux-kernel,
	Geert Uytterhoeven

If the boot CPU does not have the lowest hartid, "hartid - hbase" can
become negative, leading to an incorrect hmask, causing userspace to
crash with SEGV.  This is observed on e.g. Starlight Beta, where cpuid 1
maps to hartid 0, and cpuid 0 maps to hartid 1.

Fix this by detecting this case, and shifting the accumulated mask and
updating hbase, if possible.

Fixes: 26fb751ca37846c9 ("RISC-V: Do not use cpumask data structure for hartid bitmap")
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
---
 arch/riscv/kernel/sbi.c | 57 ++++++++++++++++++++++++++++-------------
 1 file changed, 39 insertions(+), 18 deletions(-)

diff --git a/arch/riscv/kernel/sbi.c b/arch/riscv/kernel/sbi.c
index 22444cfcd56cc646..775d3322b422fe14 100644
--- a/arch/riscv/kernel/sbi.c
+++ b/arch/riscv/kernel/sbi.c
@@ -250,7 +250,7 @@ static void __sbi_set_timer_v02(uint64_t stime_value)
 
 static int __sbi_send_ipi_v02(const struct cpumask *cpu_mask)
 {
-	unsigned long hartid, cpuid, hmask = 0, hbase = 0;
+	unsigned long hartid, cpuid, hmask = 0, hbase = 0, htop = 0;
 	struct sbiret ret = {0};
 	int result;
 
@@ -259,16 +259,27 @@ static int __sbi_send_ipi_v02(const struct cpumask *cpu_mask)
 
 	for_each_cpu(cpuid, cpu_mask) {
 		hartid = cpuid_to_hartid_map(cpuid);
-		if (hmask && ((hbase + BITS_PER_LONG) <= hartid)) {
-			ret = sbi_ecall(SBI_EXT_IPI, SBI_EXT_IPI_SEND_IPI,
-					hmask, hbase, 0, 0, 0, 0);
-			if (ret.error)
-				goto ecall_failed;
-			hmask = 0;
-			hbase = 0;
+		if (hmask) {
+			if (hartid + BITS_PER_LONG <= htop ||
+			    hbase + BITS_PER_LONG <= hartid) {
+				ret = sbi_ecall(SBI_EXT_IPI,
+						SBI_EXT_IPI_SEND_IPI, hmask,
+						hbase, 0, 0, 0, 0);
+				if (ret.error)
+					goto ecall_failed;
+				hmask = 0;
+			} else if (hartid < hbase) {
+				/* shift the mask to fit lower hartid */
+				hmask <<= hbase - hartid;
+				hbase = hartid;
+			}
 		}
-		if (!hmask)
+		if (!hmask) {
 			hbase = hartid;
+			htop = hartid;
+		} else if (hartid > htop) {
+			htop = hartid;
+		}
 		hmask |= BIT(hartid - hbase);
 	}
 
@@ -345,7 +356,7 @@ static int __sbi_rfence_v02(int fid, const struct cpumask *cpu_mask,
 			    unsigned long start, unsigned long size,
 			    unsigned long arg4, unsigned long arg5)
 {
-	unsigned long hartid, cpuid, hmask = 0, hbase = 0;
+	unsigned long hartid, cpuid, hmask = 0, hbase = 0, htop = 0;
 	int result;
 
 	if (!cpu_mask || cpumask_empty(cpu_mask))
@@ -353,16 +364,26 @@ static int __sbi_rfence_v02(int fid, const struct cpumask *cpu_mask,
 
 	for_each_cpu(cpuid, cpu_mask) {
 		hartid = cpuid_to_hartid_map(cpuid);
-		if (hmask && ((hbase + BITS_PER_LONG) <= hartid)) {
-			result = __sbi_rfence_v02_call(fid, hmask, hbase,
-						       start, size, arg4, arg5);
-			if (result)
-				return result;
-			hmask = 0;
-			hbase = 0;
+		if (hmask) {
+			if (hartid + BITS_PER_LONG <= htop ||
+			    hbase + BITS_PER_LONG <= hartid) {
+				result = __sbi_rfence_v02_call(fid, hmask,
+						hbase, start, size, arg4, arg5);
+				if (result)
+					return result;
+				hmask = 0;
+			} else if (hartid < hbase) {
+				/* shift the mask to fit lower hartid */
+				hmask <<= hbase - hartid;
+				hbase = hartid;
+			}
 		}
-		if (!hmask)
+		if (!hmask) {
 			hbase = hartid;
+			htop = hartid;
+		} else if (hartid > htop) {
+			htop = hartid;
+		}
 		hmask |= BIT(hartid - hbase);
 	}
 
-- 
2.25.1


^ permalink raw reply related	[flat|nested] 7+ messages in thread

* Re: [PATCH 0/3] RISC-V: Fix cpumask rework falloout
  2022-01-31 13:26 [PATCH 0/3] RISC-V: Fix cpumask rework falloout Geert Uytterhoeven
                   ` (2 preceding siblings ...)
  2022-01-31 13:26 ` [PATCH 3/3] RISC-V: Fix IPI/RFENCE hmask on non-monotonic hartid ordering Geert Uytterhoeven
@ 2022-01-31 17:52 ` Emil Renner Berthing
  3 siblings, 0 replies; 7+ messages in thread
From: Emil Renner Berthing @ 2022-01-31 17:52 UTC (permalink / raw)
  To: Geert Uytterhoeven
  Cc: Paul Walmsley, Palmer Dabbelt, Albert Ou, Anup Patel,
	Atish Patra, Jessica Clarke, linux-riscv,
	Linux Kernel Mailing List

On Mon, 31 Jan 2022 at 14:26, Geert Uytterhoeven <geert@linux-m68k.org> wrote:
>
>         Hi all,
>
> This patch series intends to fix the fall-out from commit
> 26fb751ca37846c9 ("RISC-V: Do not use cpumask data structure for hartid
> bitmap"), cfr. the discussion in thread[1].
>
> Thanks for your comments!
>
> [1] "[PATCH v3 6/6] RISC-V: Do not use cpumask data structure for hartid
>      bitmap"
>     https://lore.kernel.org/r/20220120090918.2646626-7-atishp@rivosinc.com/
>
> Geert Uytterhoeven (3):
>   RISC-V: Fix hartid mask handling for hartid 31 and up
>   RISC-V: Fix handling of empty cpu masks
>   RISC-V: Fix IPI/RFENCE hmask on non-monotonic hartid ordering
>
>  arch/riscv/kernel/sbi.c | 72 +++++++++++++++++++++++++++--------------
>  1 file changed, 47 insertions(+), 25 deletions(-)

I built this and rebooted my Starlight board a number of times without
seeing any of the errors that usually shows up when [1] is not
reverted.

Tested-by: Emil Renner Berthing <kernel@esmil.dk>

[1]: 26fb751ca37846c9 ("RISC-V: Do not use cpumask data structure for
hartid bitmap")

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH 2/3] RISC-V: Fix handling of empty cpu masks
  2022-01-31 13:26 ` [PATCH 2/3] RISC-V: Fix handling of empty cpu masks Geert Uytterhoeven
@ 2022-01-31 21:16   ` Atish Patra
  0 siblings, 0 replies; 7+ messages in thread
From: Atish Patra @ 2022-01-31 21:16 UTC (permalink / raw)
  To: Geert Uytterhoeven
  Cc: Paul Walmsley, Palmer Dabbelt, Albert Ou, Anup Patel,
	Atish Patra, Jessica Clarke, Emil Renner Berthing, linux-riscv,
	linux-kernel@vger.kernel.org List

On Mon, Jan 31, 2022 at 5:26 AM Geert Uytterhoeven <geert@linux-m68k.org> wrote:
>
> The cpumask rework slightly changed the behavior of the code.  Fix this
> by treating an empty cpumask as meaning all online CPUs.
>
> Extracted from a patch by Atish Patra <atishp@rivosinc.com>.
>
> Reported-by: Jessica Clarke <jrtc27@jrtc27.com>
> Fixes: 26fb751ca37846c9 ("RISC-V: Do not use cpumask data structure for hartid bitmap")
> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
> ---
>  arch/riscv/kernel/sbi.c | 8 ++++----
>  1 file changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/arch/riscv/kernel/sbi.c b/arch/riscv/kernel/sbi.c
> index f93bc75f2bc43c8e..22444cfcd56cc646 100644
> --- a/arch/riscv/kernel/sbi.c
> +++ b/arch/riscv/kernel/sbi.c
> @@ -161,7 +161,7 @@ static int __sbi_send_ipi_v01(const struct cpumask *cpu_mask)
>  {
>         unsigned long hart_mask;
>
> -       if (!cpu_mask)
> +       if (!cpu_mask || cpumask_empty(cpu_mask))
>                 cpu_mask = cpu_online_mask;
>         hart_mask = __sbi_v01_cpumask_to_hartmask(cpu_mask);
>
> @@ -177,7 +177,7 @@ static int __sbi_rfence_v01(int fid, const struct cpumask *cpu_mask,
>         int result = 0;
>         unsigned long hart_mask;
>
> -       if (!cpu_mask)
> +       if (!cpu_mask || cpumask_empty(cpu_mask))
>                 cpu_mask = cpu_online_mask;
>         hart_mask = __sbi_v01_cpumask_to_hartmask(cpu_mask);
>
> @@ -254,7 +254,7 @@ static int __sbi_send_ipi_v02(const struct cpumask *cpu_mask)
>         struct sbiret ret = {0};
>         int result;
>
> -       if (!cpu_mask)
> +       if (!cpu_mask || cpumask_empty(cpu_mask))
>                 cpu_mask = cpu_online_mask;
>
>         for_each_cpu(cpuid, cpu_mask) {
> @@ -348,7 +348,7 @@ static int __sbi_rfence_v02(int fid, const struct cpumask *cpu_mask,
>         unsigned long hartid, cpuid, hmask = 0, hbase = 0;
>         int result;
>
> -       if (!cpu_mask)
> +       if (!cpu_mask || cpumask_empty(cpu_mask))
>                 cpu_mask = cpu_online_mask;
>
>         for_each_cpu(cpuid, cpu_mask) {
> --
> 2.25.1
>
>
> _______________________________________________
> linux-riscv mailing list
> linux-riscv@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-riscv


Reviewed-by: Atish Patra <atishp@rivosinc.com>


-- 
Regards,
Atish

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH 3/3] RISC-V: Fix IPI/RFENCE hmask on non-monotonic hartid ordering
  2022-01-31 13:26 ` [PATCH 3/3] RISC-V: Fix IPI/RFENCE hmask on non-monotonic hartid ordering Geert Uytterhoeven
@ 2022-01-31 21:18   ` Atish Patra
  0 siblings, 0 replies; 7+ messages in thread
From: Atish Patra @ 2022-01-31 21:18 UTC (permalink / raw)
  To: Geert Uytterhoeven
  Cc: Paul Walmsley, Palmer Dabbelt, Albert Ou, Anup Patel,
	Atish Patra, Jessica Clarke, Emil Renner Berthing, linux-riscv,
	linux-kernel@vger.kernel.org List

On Mon, Jan 31, 2022 at 5:26 AM Geert Uytterhoeven <geert@linux-m68k.org> wrote:
>
> If the boot CPU does not have the lowest hartid, "hartid - hbase" can
> become negative, leading to an incorrect hmask, causing userspace to
> crash with SEGV.  This is observed on e.g. Starlight Beta, where cpuid 1
> maps to hartid 0, and cpuid 0 maps to hartid 1.
>
> Fix this by detecting this case, and shifting the accumulated mask and
> updating hbase, if possible.
>
> Fixes: 26fb751ca37846c9 ("RISC-V: Do not use cpumask data structure for hartid bitmap")
> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
> ---
>  arch/riscv/kernel/sbi.c | 57 ++++++++++++++++++++++++++++-------------
>  1 file changed, 39 insertions(+), 18 deletions(-)
>
> diff --git a/arch/riscv/kernel/sbi.c b/arch/riscv/kernel/sbi.c
> index 22444cfcd56cc646..775d3322b422fe14 100644
> --- a/arch/riscv/kernel/sbi.c
> +++ b/arch/riscv/kernel/sbi.c
> @@ -250,7 +250,7 @@ static void __sbi_set_timer_v02(uint64_t stime_value)
>
>  static int __sbi_send_ipi_v02(const struct cpumask *cpu_mask)
>  {
> -       unsigned long hartid, cpuid, hmask = 0, hbase = 0;
> +       unsigned long hartid, cpuid, hmask = 0, hbase = 0, htop = 0;
>         struct sbiret ret = {0};
>         int result;
>
> @@ -259,16 +259,27 @@ static int __sbi_send_ipi_v02(const struct cpumask *cpu_mask)
>
>         for_each_cpu(cpuid, cpu_mask) {
>                 hartid = cpuid_to_hartid_map(cpuid);
> -               if (hmask && ((hbase + BITS_PER_LONG) <= hartid)) {
> -                       ret = sbi_ecall(SBI_EXT_IPI, SBI_EXT_IPI_SEND_IPI,
> -                                       hmask, hbase, 0, 0, 0, 0);
> -                       if (ret.error)
> -                               goto ecall_failed;
> -                       hmask = 0;
> -                       hbase = 0;
> +               if (hmask) {
> +                       if (hartid + BITS_PER_LONG <= htop ||
> +                           hbase + BITS_PER_LONG <= hartid) {
> +                               ret = sbi_ecall(SBI_EXT_IPI,
> +                                               SBI_EXT_IPI_SEND_IPI, hmask,
> +                                               hbase, 0, 0, 0, 0);
> +                               if (ret.error)
> +                                       goto ecall_failed;
> +                               hmask = 0;
> +                       } else if (hartid < hbase) {
> +                               /* shift the mask to fit lower hartid */
> +                               hmask <<= hbase - hartid;
> +                               hbase = hartid;
> +                       }
>                 }
> -               if (!hmask)
> +               if (!hmask) {
>                         hbase = hartid;
> +                       htop = hartid;
> +               } else if (hartid > htop) {
> +                       htop = hartid;
> +               }
>                 hmask |= BIT(hartid - hbase);
>         }
>
> @@ -345,7 +356,7 @@ static int __sbi_rfence_v02(int fid, const struct cpumask *cpu_mask,
>                             unsigned long start, unsigned long size,
>                             unsigned long arg4, unsigned long arg5)
>  {
> -       unsigned long hartid, cpuid, hmask = 0, hbase = 0;
> +       unsigned long hartid, cpuid, hmask = 0, hbase = 0, htop = 0;
>         int result;
>
>         if (!cpu_mask || cpumask_empty(cpu_mask))
> @@ -353,16 +364,26 @@ static int __sbi_rfence_v02(int fid, const struct cpumask *cpu_mask,
>
>         for_each_cpu(cpuid, cpu_mask) {
>                 hartid = cpuid_to_hartid_map(cpuid);
> -               if (hmask && ((hbase + BITS_PER_LONG) <= hartid)) {
> -                       result = __sbi_rfence_v02_call(fid, hmask, hbase,
> -                                                      start, size, arg4, arg5);
> -                       if (result)
> -                               return result;
> -                       hmask = 0;
> -                       hbase = 0;
> +               if (hmask) {
> +                       if (hartid + BITS_PER_LONG <= htop ||
> +                           hbase + BITS_PER_LONG <= hartid) {
> +                               result = __sbi_rfence_v02_call(fid, hmask,
> +                                               hbase, start, size, arg4, arg5);
> +                               if (result)
> +                                       return result;
> +                               hmask = 0;
> +                       } else if (hartid < hbase) {
> +                               /* shift the mask to fit lower hartid */
> +                               hmask <<= hbase - hartid;
> +                               hbase = hartid;
> +                       }
>                 }
> -               if (!hmask)
> +               if (!hmask) {
>                         hbase = hartid;
> +                       htop = hartid;
> +               } else if (hartid > htop) {
> +                       htop = hartid;
> +               }
>                 hmask |= BIT(hartid - hbase);
>         }
>
> --
> 2.25.1
>
>
> _______________________________________________
> linux-riscv mailing list
> linux-riscv@lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/linux-riscv


Reviewed-by: Atish Patra <atishp@rivosinc.com>
Tested-by: Atish Patra <atishp@rivosinc.com>

-- 
Regards,
Atish

^ permalink raw reply	[flat|nested] 7+ messages in thread

end of thread, other threads:[~2022-01-31 21:18 UTC | newest]

Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-01-31 13:26 [PATCH 0/3] RISC-V: Fix cpumask rework falloout Geert Uytterhoeven
2022-01-31 13:26 ` [PATCH 1/3] RISC-V: Fix hartid mask handling for hartid 31 and up Geert Uytterhoeven
2022-01-31 13:26 ` [PATCH 2/3] RISC-V: Fix handling of empty cpu masks Geert Uytterhoeven
2022-01-31 21:16   ` Atish Patra
2022-01-31 13:26 ` [PATCH 3/3] RISC-V: Fix IPI/RFENCE hmask on non-monotonic hartid ordering Geert Uytterhoeven
2022-01-31 21:18   ` Atish Patra
2022-01-31 17:52 ` [PATCH 0/3] RISC-V: Fix cpumask rework falloout Emil Renner Berthing

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).