Linux-MIPS Archive mirror
 help / color / mirror / Atom feed
From: kpursoty@proton.me
To: "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"daniel.lezcano@linaro.org" <daniel.lezcano@linaro.org>,
	"tglx@linutronix.de" <tglx@linutronix.de>
Cc: "linux-mips@vger.kernel.org" <linux-mips@vger.kernel.org>,
	"tsbogend@alpha.franken.de" <tsbogend@alpha.franken.de>,
	"cjd@cjdns.fr" <cjd@cjdns.fr>,
	"naseefkm@gmail.com" <naseefkm@gmail.com>
Subject: Re: [PATCH v3 1/2] clocksource: timer-econet-en751221: fix timer block mapping at boot
Date: Thu, 16 Apr 2026 07:19:45 +0000	[thread overview]
Message-ID: <WYwhiNQWwEXQoD7-dVONf2PgvQQR5adVWEcJzlAcgyF2XcoXEw3atvsPu5qiNzGPbu_rT6i6OD5Utc4L_dCxwmkxRQ5ugS4wOMH1G80JyNw=@proton.me> (raw)
In-Reply-To: <o_cMD_cD6PNolsucCVtoojfSsfFgjsNiEop7YJ_QW8jrX1aCkIF8Rhdk-8_DSCBU1WrKSjr9STxTuCjiRK8yJ6vDA4Q1Ck2BCgOzYaBqSEs=@proton.me>

Thanks for your notes. Patch withdrawn.

On Wednesday, 15 April 2026 at 08:53, kpursoty@proton.me <kpursoty@proton.me> wrote:

> timer_init() used DIV_ROUND_UP(num_possible_cpus(), 2) to determine how
> many register blocks to iomap. At early boot with VPE-based SMP, MIPS
> reports num_possible_cpus()=1 (VPEs not yet brought online), giving
> num_blocks=1. Only membase[0] is then mapped via of_iomap.
>
> The EN751627 SoC has 2 physical cores, each with 2 VPEs, giving NR_CPUS=4
> and two timer register blocks (one per core). Each block serves two VPEs:
> block 0 handles CPU0+CPU1 (core 0), block 1 handles CPU2+CPU3 (core 1).
> The block count is a silicon constant: DIV_ROUND_UP(NR_CPUS, 2) = 2.
>
> cevt_init() calls cevt_dev_init(i) for each possible CPU via
> for_each_possible_cpu(). On EN7528/EN751627 with 4 VPEs, NR_CPUS=4 so
> cevt_dev_init(2) is called. cevt_dev_init(2) writes to reg_compare(2)
> which dereferences membase[2>>1] = membase[1], which is NULL:
>
>   CPU 0 Unable to handle kernel paging request at virtual address 00000008
>   epc : iowrite32+0x4/0x10
>   ra  : cevt_dev_init+0x40/0x64
>
> Fix: replace the runtime calculation with ECONET_NUM_BLOCKS, which is
> DIV_ROUND_UP(NR_CPUS, 2) evaluated at compile time. This is the same
> expression used to declare the membase[] array, so the loop bound and
> array size are provably consistent. For NR_CPUS=4 this is always 2,
> correctly mapping both register blocks regardless of how many VPEs are
> visible at early boot.
>
> Fixes: 3b4c33ac87d0 ("clocksource/drivers: Add EcoNet Timer HPT driver")
> Signed-off-by: Kervin Pursoty <kpursoty@proton.me>
> Reviewed-by: Caleb James DeLisle <cjd@cjdns.fr>
> ---
> v3: No changes. Resent as v3 to keep version in sync with 2/2.
> v2: Add Reviewed-by from Caleb James DeLisle.
>
>  drivers/clocksource/timer-econet-en751221.c | 3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
>
> diff --git a/drivers/clocksource/timer-econet-en751221.c b/drivers/clocksource/timer-econet-en751221.c
> --- a/drivers/clocksource/timer-econet-en751221.c
> +++ b/drivers/clocksource/timer-econet-en751221.c
> @@ -160,5 +160,4 @@ static u64 notrace sched_clock_read(void)
>  static int __init timer_init(struct device_node *np)
>  {
> -	int num_blocks = DIV_ROUND_UP(num_possible_cpus(), 2);
>  	struct clk *clk;
>  	int ret;
>
> @@ -172,5 +171,5 @@ static int __init timer_init(struct device_node *np)
>
>  	econet_timer.freq_hz = clk_get_rate(clk);
>
> -	for (int i = 0; i < num_blocks; i++) {
> +	for (int i = 0; i < ECONET_NUM_BLOCKS; i++) {
>  		econet_timer.membase[i] = of_iomap(np, i);
>  		if (!econet_timer.membase[i]) {
>  			pr_err("%pOFn: failed to map register [%d]\n", np, i);

      reply	other threads:[~2026-04-16  7:19 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-04-15  7:53 [PATCH v3 1/2] clocksource: timer-econet-en751221: fix timer block mapping at boot kpursoty
2026-04-16  7:19 ` kpursoty [this message]

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='WYwhiNQWwEXQoD7-dVONf2PgvQQR5adVWEcJzlAcgyF2XcoXEw3atvsPu5qiNzGPbu_rT6i6OD5Utc4L_dCxwmkxRQ5ugS4wOMH1G80JyNw=@proton.me' \
    --to=kpursoty@proton.me \
    --cc=cjd@cjdns.fr \
    --cc=daniel.lezcano@linaro.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mips@vger.kernel.org \
    --cc=naseefkm@gmail.com \
    --cc=tglx@linutronix.de \
    --cc=tsbogend@alpha.franken.de \
    /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 read-only IMAP folder(s) and NNTP newsgroup(s).