From: Bjorn Andersson <bjorn.andersson@linaro.org> To: Peng Fan <peng.fan@nxp.com> Cc: ohad@wizery.com, linux-remoteproc@vger.kernel.org, linux-kernel@vger.kernel.org, linux-imx@nxp.com Subject: Re: [PATCH 1/2] remoteproc: drop memset when loading elf segments Date: Thu, 9 Apr 2020 18:20:37 -0700 [thread overview] Message-ID: <20200410012034.GU20625@builder.lan> (raw) In-Reply-To: <1586420572-28353-1-git-send-email-peng.fan@nxp.com> On Thu 09 Apr 01:22 PDT 2020, Peng Fan wrote: > To arm64, "dc zva, dst" is used in memset. > Per ARM DDI 0487A.j, chapter C5.3.8 DC ZVA, Data Cache Zero by VA, > > "If the memory region being zeroed is any type of Device memory, > this instruction can give an alignment fault which is prioritized > in the same way as other alignment faults that are determined > by the memory type." > > On i.MX platforms, when elf is loaded to onchip TCM area, the region > is ioremapped, so "dc zva, dst" will trigger abort. > > Since memset is not strictly required, let's drop it. > This would imply that we trust that the firmware doesn't expect remoteproc to zero out the memory, which we've always done. So I don't think we can say that it's not required. > Signed-off-by: Peng Fan <peng.fan@nxp.com> > --- > drivers/remoteproc/remoteproc_elf_loader.c | 7 ++----- > 1 file changed, 2 insertions(+), 5 deletions(-) > > diff --git a/drivers/remoteproc/remoteproc_elf_loader.c b/drivers/remoteproc/remoteproc_elf_loader.c > index 16e2c496fd45..cc50fe70d50c 100644 > --- a/drivers/remoteproc/remoteproc_elf_loader.c > +++ b/drivers/remoteproc/remoteproc_elf_loader.c > @@ -238,14 +238,11 @@ int rproc_elf_load_segments(struct rproc *rproc, const struct firmware *fw) > memcpy(ptr, elf_data + offset, filesz); > > /* > - * Zero out remaining memory for this segment. > + * No need zero out remaining memory for this segment. > * > * This isn't strictly required since dma_alloc_coherent already > - * did this for us. albeit harmless, we may consider removing > - * this. > + * did this for us. In the case of recovery this comment is wrong, we do not dma_alloc_coherent() the carveout during a recovery. And in your case you ioremapped existing TCM, so it's never true. > */ > - if (memsz > filesz) > - memset(ptr + filesz, 0, memsz - filesz); So I think you do want to zero out this region. Question is how we do it... Regards, Bjorn > } > > if (ret == 0) > -- > 2.16.4 >
WARNING: multiple messages have this Message-ID (diff)
From: Bjorn Andersson <bjorn.andersson@linaro.org> To: Peng Fan <peng.fan@nxp.com> Cc: ohad@wizery.com, linux-remoteproc@vger.kernel.org, linux-kernel@vger.kernel.org, linux-imx@nxp.com Subject: Re: [PATCH 1/2] remoteproc: drop memset when loading elf segments Date: Thu, 9 Apr 2020 18:20:34 -0700 [thread overview] Message-ID: <20200410012034.GU20625@builder.lan> (raw) Message-ID: <20200410012034.odwthBqUmGyYLdPJPW5Wbj6oJSe-4eK-j3Xjk9c4-uc@z> (raw) In-Reply-To: <1586420572-28353-1-git-send-email-peng.fan@nxp.com> On Thu 09 Apr 01:22 PDT 2020, Peng Fan wrote: > To arm64, "dc zva, dst" is used in memset. > Per ARM DDI 0487A.j, chapter C5.3.8 DC ZVA, Data Cache Zero by VA, > > "If the memory region being zeroed is any type of Device memory, > this instruction can give an alignment fault which is prioritized > in the same way as other alignment faults that are determined > by the memory type." > > On i.MX platforms, when elf is loaded to onchip TCM area, the region > is ioremapped, so "dc zva, dst" will trigger abort. > > Since memset is not strictly required, let's drop it. > This would imply that we trust that the firmware doesn't expect remoteproc to zero out the memory, which we've always done. So I don't think we can say that it's not required. > Signed-off-by: Peng Fan <peng.fan@nxp.com> > --- > drivers/remoteproc/remoteproc_elf_loader.c | 7 ++----- > 1 file changed, 2 insertions(+), 5 deletions(-) > > diff --git a/drivers/remoteproc/remoteproc_elf_loader.c b/drivers/remoteproc/remoteproc_elf_loader.c > index 16e2c496fd45..cc50fe70d50c 100644 > --- a/drivers/remoteproc/remoteproc_elf_loader.c > +++ b/drivers/remoteproc/remoteproc_elf_loader.c > @@ -238,14 +238,11 @@ int rproc_elf_load_segments(struct rproc *rproc, const struct firmware *fw) > memcpy(ptr, elf_data + offset, filesz); > > /* > - * Zero out remaining memory for this segment. > + * No need zero out remaining memory for this segment. > * > * This isn't strictly required since dma_alloc_coherent already > - * did this for us. albeit harmless, we may consider removing > - * this. > + * did this for us. In the case of recovery this comment is wrong, we do not dma_alloc_coherent() the carveout during a recovery. And in your case you ioremapped existing TCM, so it's never true. > */ > - if (memsz > filesz) > - memset(ptr + filesz, 0, memsz - filesz); So I think you do want to zero out this region. Question is how we do it... Regards, Bjorn > } > > if (ret == 0) > -- > 2.16.4 >
next prev parent reply other threads:[~2020-04-10 1:20 UTC|newest] Thread overview: 29+ messages / expand[flat|nested] mbox.gz Atom feed top 2020-04-09 8:22 [PATCH 1/2] remoteproc: drop memset when loading elf segments Peng Fan 2020-04-09 8:22 ` Peng Fan 2020-04-09 8:22 ` [PATCH 2/2] remoteproc: use filesz as backup when translate memsz fail Peng Fan 2020-04-09 8:22 ` Peng Fan 2020-04-10 1:22 ` Bjorn Andersson 2020-04-10 1:22 ` Bjorn Andersson 2020-04-10 1:32 ` Peng Fan 2020-04-10 1:32 ` Peng Fan 2020-04-11 1:55 ` Bjorn Andersson 2020-04-11 1:55 ` Bjorn Andersson 2020-04-17 19:21 ` Mathieu Poirier 2020-04-17 19:21 ` Mathieu Poirier 2020-04-18 9:10 ` Peng Fan 2020-04-18 9:10 ` Peng Fan 2020-04-10 1:20 ` Bjorn Andersson [this message] 2020-04-10 1:20 ` [PATCH 1/2] remoteproc: drop memset when loading elf segments Bjorn Andersson 2020-04-10 1:29 ` Peng Fan 2020-04-10 1:29 ` Peng Fan 2020-04-11 1:51 ` Bjorn Andersson 2020-04-11 1:51 ` Bjorn Andersson 2020-04-13 9:05 ` Peng Fan 2020-04-13 9:05 ` Peng Fan 2020-04-17 16:43 ` Suman Anna 2020-04-17 16:43 ` Suman Anna 2020-04-21 7:42 ` Peng Fan 2020-04-21 7:42 ` Peng Fan 2020-04-21 18:25 ` Suman Anna 2020-04-21 18:25 ` Suman Anna 2020-05-11 9:15 ` Clément Leger
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=20200410012034.GU20625@builder.lan \ --to=bjorn.andersson@linaro.org \ --cc=linux-imx@nxp.com \ --cc=linux-kernel@vger.kernel.org \ --cc=linux-remoteproc@vger.kernel.org \ --cc=ohad@wizery.com \ --cc=peng.fan@nxp.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: linkBe 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).