From: Bjorn Andersson <bjorn.andersson@linaro.org>
To: Loic Pallardy <loic.pallardy@st.com>
Cc: ohad@wizery.com, linux-remoteproc@vger.kernel.org,
linux-kernel@vger.kernel.org, arnaud.pouliquen@st.com,
benjamin.gaignard@linaro.org
Subject: Re: [PATCH v2 05/16] remoteproc: modify rproc_handle_carveout to support preallocated region
Date: Wed, 13 Dec 2017 16:59:17 -0800 [thread overview]
Message-ID: <20171214005917.GG17344@builder> (raw)
In-Reply-To: <1512060411-729-6-git-send-email-loic.pallardy@st.com>
On Thu 30 Nov 08:46 PST 2017, Loic Pallardy wrote:
> In current version rproc_handle_carveout function support only dynamic
> region allocation.
> This patch extends rproc_handle_carveout function to support different carveout
> configurations:
> - fixed DA and fixed PA: check if already part of pre-registered carveouts
> (platform driver). If no, return error.
> - fixed DA and any PA: check if already part of pre-allocated carveouts
> (platform driver). If not found and rproc supports iommu, continue with
> dynamic allocation (DA will be used for iommu programming), else return
> error as no way to force DA.
> - any DA and any PA: use original dynamic allocation
>
> Signed-off-by: Loic Pallardy <loic.pallardy@st.com>
> ---
> drivers/remoteproc/remoteproc_core.c | 40 ++++++++++++++++++++++++++++++++++++
> 1 file changed, 40 insertions(+)
>
> diff --git a/drivers/remoteproc/remoteproc_core.c b/drivers/remoteproc/remoteproc_core.c
> index 78525d1..515a17a 100644
> --- a/drivers/remoteproc/remoteproc_core.c
> +++ b/drivers/remoteproc/remoteproc_core.c
> @@ -184,6 +184,10 @@ void *rproc_da_to_va(struct rproc *rproc, u64 da, int len)
> struct rproc_mem_entry *carveout;
> void *ptr = NULL;
>
> + /*
> + * da_to_va platform driver is deprecated. Driver should register
> + * carveout thanks to rproc_add_carveout function
> + */
I think this comment is unrelated to the rest of this patch. I also
think that at the end of the carveout-rework we should have a patch
removing this ops.
> if (rproc->ops->da_to_va) {
> ptr = rproc->ops->da_to_va(rproc, da, len);
> if (ptr)
> @@ -677,6 +681,7 @@ static int rproc_handle_carveout(struct rproc *rproc,
> struct rproc_mem_entry *carveout, *mapping;
> struct device *dev = &rproc->dev;
> dma_addr_t dma;
> + phys_addr_t pa;
> void *va;
> int ret;
>
> @@ -698,6 +703,41 @@ static int rproc_handle_carveout(struct rproc *rproc,
> if (!carveout)
> return -ENOMEM;
>
> + /* Check carveout rsc already part of a registered carveout */
> + if (rsc->da != FW_RSC_ADDR_ANY) {
As mentioned before, I consider it perfectly viable for rsc->da to be
ANY and the driver providing a fixed carveout.
> + va = rproc_find_carveout_by_da(rproc, rsc->da, rsc->len);
> +
> + if (va) {
In a system with an iommu it's possible that rsc->len is larger than
some carveout->len and va is NULL here so we fall through, allocate some
memory and remap a segment of the carveout. (Or hopefully fails
attempting).
> + /* Registered region found */
> + pa = rproc_va_to_pa(va);
> + if (rsc->pa != FW_RSC_ADDR_ANY && rsc->pa != (u32)pa) {
> + /* Carveout doesn't match request */
> + dev_err(dev->parent,
> + "Failed to find carveout fitting da and pa\n");
> + return -ENOMEM;
> + }
> +
> + /* Update rsc table with physical address */
> + rsc->pa = (u32)pa;
> +
> + /* Update carveouts list */
> + carveout->va = va;
> + carveout->len = rsc->len;
> + carveout->da = rsc->da;
> + carveout->priv = (void *)CARVEOUT_RSC;
> +
> + list_add_tail(&carveout->node, &rproc->carveouts);
rproc_find_carveout_by_da() will return a reference into a carveout, now
we add another overlapping carveout into the same list.
I think it would be saner to not allow the resource table to describe
subsets of carveouts registered by the driver.
In which case this would better find a carveout by name or exact da,
then check that the pa, da, len and rsc->flags are adequate.
> +
> + return 0;
> + }
> +
> + if (!rproc->domain) {
Currently this function ignore invalid values of da when !domain, so I
think it would be good you can submit this sanity check in it's own
patch so that anyone bisecting this would know why their broken firmware
suddenly isn't loadable.
> + dev_err(dev->parent,
> + "Bad carveout rsc configuration\n");
> + return -ENOMEM;
> + }
> + }
> +
Regards,
Bjorn
next prev parent reply other threads:[~2017-12-14 0:59 UTC|newest]
Thread overview: 48+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-11-30 16:46 [PATCH v2 00/16] remoteproc: add fixed memory region support Loic Pallardy
2017-11-30 16:46 ` [PATCH v2 01/16] remoteproc: add rproc_va_to_pa function Loic Pallardy
2017-12-14 0:30 ` Bjorn Andersson
2018-01-12 7:43 ` Loic PALLARDY
2017-11-30 16:46 ` [PATCH v2 02/16] remoteproc: add release ops in rproc_mem_entry struct Loic Pallardy
2017-12-14 0:34 ` Bjorn Andersson
2018-01-12 7:43 ` Loic PALLARDY
2017-11-30 16:46 ` [PATCH v2 03/16] remoteproc: introduce rproc_add_carveout function Loic Pallardy
2017-12-14 0:36 ` Bjorn Andersson
2018-01-12 7:45 ` Loic PALLARDY
2017-11-30 16:46 ` [PATCH v2 04/16] remoteproc: introduce rproc_find_carveout_by_da Loic Pallardy
2017-12-14 0:45 ` Bjorn Andersson
2018-01-12 7:48 ` Loic PALLARDY
2017-11-30 16:46 ` [PATCH v2 05/16] remoteproc: modify rproc_handle_carveout to support preallocated region Loic Pallardy
2017-12-14 0:59 ` Bjorn Andersson [this message]
2018-01-12 7:56 ` Loic PALLARDY
2018-10-23 17:40 ` Suman Anna
2018-10-23 19:09 ` Loic PALLARDY
2018-10-23 19:12 ` Suman Anna
2017-11-30 16:46 ` [PATCH v2 06/16] remoteproc: modify vring allocation " Loic Pallardy
2017-12-14 1:09 ` Bjorn Andersson
2018-01-12 8:13 ` Loic PALLARDY
2017-11-30 16:46 ` [PATCH v2 07/16] remoteproc: st: add reserved memory support Loic Pallardy
2017-12-14 1:15 ` Bjorn Andersson
2018-01-12 8:19 ` Loic PALLARDY
2017-11-30 16:46 ` [PATCH v2 08/16] remoteproc: add name in rproc_mem_entry struct Loic Pallardy
2017-12-14 1:21 ` Bjorn Andersson
2018-01-12 8:19 ` Loic PALLARDY
2017-11-30 16:46 ` [PATCH v2 09/16] remoteproc: add memory device management support Loic Pallardy
2017-11-30 16:46 ` [PATCH v2 10/16] remoteproc: add memory device registering in rproc_add_carveout Loic Pallardy
2017-12-14 1:29 ` Bjorn Andersson
2018-01-15 9:09 ` Loic PALLARDY
2017-11-30 16:46 ` [PATCH v2 11/16] remoteproc: introduce rproc_find_carveout_by_name function Loic Pallardy
2017-12-14 1:32 ` Bjorn Andersson
2018-01-15 9:10 ` Loic PALLARDY
2017-11-30 16:46 ` [PATCH v2 12/16] remoteproc: look-up memory-device for vring allocation Loic Pallardy
2017-12-14 1:44 ` Bjorn Andersson
2018-01-15 20:44 ` Loic PALLARDY
2017-11-30 16:46 ` [PATCH v2 13/16] remoteproc: look-up memory-device for virtio device allocation Loic Pallardy
2017-12-14 5:32 ` Bjorn Andersson
2018-01-15 20:57 ` Loic PALLARDY
2017-11-30 16:46 ` [PATCH v2 14/16] remoteproc: look-up pre-registered carveout by name for carveout allocation Loic Pallardy
2017-12-14 5:34 ` Bjorn Andersson
2018-01-15 20:59 ` Loic PALLARDY
2017-11-30 16:46 ` [PATCH v2 15/16] remoteproc: st: associate memory device to memory regions Loic Pallardy
2017-12-14 5:37 ` Bjorn Andersson
2018-01-15 21:04 ` Loic PALLARDY
2017-11-30 16:46 ` [PATCH v2 16/16] rpmsg: virtio: allocate buffer from parent Loic Pallardy
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=20171214005917.GG17344@builder \
--to=bjorn.andersson@linaro.org \
--cc=arnaud.pouliquen@st.com \
--cc=benjamin.gaignard@linaro.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-remoteproc@vger.kernel.org \
--cc=loic.pallardy@st.com \
--cc=ohad@wizery.com \
/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).