From patchwork Wed Dec 13 10:49:17 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Philipp Stanner X-Patchwork-Id: 13490669 Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=redhat.com header.i=@redhat.com header.b="RhlJT0Pv" Received: from us-smtp-delivery-124.mimecast.com (us-smtp-delivery-124.mimecast.com [170.10.133.124]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 6D091E4 for ; Wed, 13 Dec 2023 02:49:32 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1702464571; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:cc:mime-version:mime-version: content-transfer-encoding:content-transfer-encoding; bh=Jsenl4ryPejHUAvVnk0GRNLRuch6fMe21jriRXS4vNY=; b=RhlJT0PvLGVVp0hI20MxX8lTSj2OwdZ/tT4k9Vsg6H5DfKawbQzlVOmo9fKRe5JJxoOFiD OUHpUR0mTVn+a9hTYRgm3IfRi2scMptAoqXfUyPL0Yc7uRx4pEat6wuaTpVMW7S4iv1X7h 4UoItnA864AirPOGX0lBSvOiJhiLAlM= Received: from mail-ej1-f69.google.com (mail-ej1-f69.google.com [209.85.218.69]) by relay.mimecast.com with ESMTP with STARTTLS (version=TLSv1.3, cipher=TLS_AES_256_GCM_SHA384) id us-mta-594-6PvlrCBINsKO_qV0LHGPqw-1; Wed, 13 Dec 2023 05:49:30 -0500 X-MC-Unique: 6PvlrCBINsKO_qV0LHGPqw-1 Received: by mail-ej1-f69.google.com with SMTP id a640c23a62f3a-a1d27c45705so141560366b.0 for ; Wed, 13 Dec 2023 02:49:29 -0800 (PST) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1702464569; x=1703069369; h=content-transfer-encoding:mime-version:message-id:date:subject:cc :to:from:x-gm-message-state:from:to:cc:subject:date:message-id :reply-to; bh=Jsenl4ryPejHUAvVnk0GRNLRuch6fMe21jriRXS4vNY=; b=Q41mz6qIiAH6F+SzapBKpCHzmTWLzB8NUc2IZ8cVv1mA07DQ5CIeVSBiM/Ed/euEB9 yeg2jYblBJX+tmrCTLud87paTPbO70jFCH7Q1o3mRTsxssQkOrrlmpT4KMXkT2W+CYfa DZe36uDrWy5ubRCM8rbz2Q8mTAeQB65aqkF53/UDaWP53flubi5v9g87n8TAnr34ckGq KF4jrLey5rfuVSWpbBlIq7V2cdc2t6F17E998uzfNj9kJcKVEkKxXFLGVQ1JCV57ecDT DPK+G5iXyWo4WWwn3o9bFi5V3DqpgayMenaIvxdof5XI4VNS+rtu63hgoiRwVi/n8pYE j8kQ== X-Gm-Message-State: AOJu0YzbfMIgWlCJdyQbCeQKAfuMae9lbCeku2RkVKAirBuEZo7PQemr o/PFC/mmU6LsYaXM5Yb7vy4jijovdazn/IqmQnsVLkGtXgD1UrMbvfjuPufP1KZgqYDaBqvyP+d Zbk+rYnA2nt07XSJNezfT X-Received: by 2002:a17:906:99cf:b0:a1c:5944:29bb with SMTP id s15-20020a17090699cf00b00a1c594429bbmr8372954ejn.7.1702464568901; Wed, 13 Dec 2023 02:49:28 -0800 (PST) X-Google-Smtp-Source: AGHT+IFjgYBJbvfH0mH+1Rv7rE+9uQlzPl9IwjMtgaG3uNiiAi7RPtxxtWG9rSZgClhoFT2vsB4Rqw== X-Received: by 2002:a17:906:99cf:b0:a1c:5944:29bb with SMTP id s15-20020a17090699cf00b00a1c594429bbmr8372945ejn.7.1702464568536; Wed, 13 Dec 2023 02:49:28 -0800 (PST) Received: from pstanner-thinkpadt14sgen1.remote.csb ([2a01:599:914:ed27:4fa9:dbce:10f5:d0b9]) by smtp.gmail.com with ESMTPSA id vu8-20020a170907a64800b00a1d5c52d628sm7527135ejc.3.2023.12.13.02.49.25 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Wed, 13 Dec 2023 02:49:28 -0800 (PST) From: Philipp Stanner To: Bjorn Helgaas , Arnd Bergmann , Johannes Berg , Randy Dunlap , NeilBrown , John Sanpe , Kent Overstreet , Niklas Schnelle , Philipp Stanner , Dave Jiang , Uladzislau Koshchanka , "Masami Hiramatsu (Google)" , David Gow , Kees Cook , Rae Moar , Geert Uytterhoeven , "wuqiang.matt" , Yury Norov , Jason Baron , Thomas Gleixner , Marco Elver , Andrew Morton , Ben Dooks , dakr@redhat.com Cc: linux-kernel@vger.kernel.org, linux-pci@vger.kernel.org, linux-arch@vger.kernel.org, stable@vger.kernel.org Subject: [PATCH v5 0/5] Regather scattered PCI-Code Date: Wed, 13 Dec 2023 11:49:17 +0100 Message-ID: <20231213104922.13894-1-pstanner@redhat.com> X-Mailer: git-send-email 2.43.0 Precedence: bulk X-Mailing-List: linux-pci@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 @Stable-Kernel: You receive this patch series because its first patch fixes leaks in PCI. Changes in v5: - Add forgotten update to MAINTAINERS file. Changes in v4: - Apply Arnd's Reviewed-by's - Add ifdef CONFIG_HAS_IOPORT_MAP guard in drivers/pci/iomap.c (build error on openrisc) - Fix typo in patch no.5 Changes in v3: - Create a separate patch for the leaks in lib/iomap.c. Make it the series' first patch. (Arnd) - Turns out the aforementioned bug wasn't just accidentally removing iounmap() with the ifdef, it was also missing ioport_unmap() to begin with. Add it. - Move the ARCH_WANTS_GENERIC_IOMEM_IS_IOPORT-mechanism from asm-generic/io.h to asm-generic/ioport.h. (Arnd) - Adjust the implementation of iomem_is_ioport() in asm-generic/io.h so that it matches exactly what pci_iounmap() previously did in lib/pci_iomap.c. (Arnd) - Move the CONFIG_HAS_IOPORT guard in asm-generic/io.h so that iomem_is_ioport() will always be compiled and just returns false if there are no ports. - Add TODOs to several places informing about the generic iomem_is_ioport() in lib/iomap.c not being generic. - Add TODO about the followup work to make drivers/pci/iomap.c's pci_iounmap() actually generic. Changes in v2: - Replace patch 4, previously extending the comment about pci_iounmap() in lib/iomap.c, with a patch that moves pci_iounmap() from that file to drivers/pci/iomap.c, creating a unified version there. (Arnd) - Implement iomem_is_ioport() as a new helper in asm-generic/io.h and lib/iomap.c. (Arnd) - Move the build rule in drivers/pci/Makefile for iomap.o under the guard of #if PCI. This had to be done because when just checking for GENERIC_PCI_IOMAP being defined, the functions don't disappear, which was the case previously in lib/pci_iomap.c, where the entire file was made empty if PCI was not set by the guard #ifdef PCI. (Intel's Bots) - Rephares all patches' commit messages a little bit. Sooooooooo. I reworked v1. Please review this carefully, the IO-Ranges are obviously a bit tricky, as is the build-system / ifdef-ery. Arnd has suggested that architectures defining a custom inb() need their own iomem_is_ioport(), as well. I've grepped for inb() and found the following list of archs that define their own: - alpha - arm - m68k <-- - parisc - powerpc - sh - sparc - x86 <-- All of those have their own definitons of pci_iounmap(). Therefore, they don't need our generic version in the first place and, thus, also need no iomem_is_ioport(). The two exceptions are x86 and m68k. The former uses lib/iomap.c through CONFIG_GENERIC_IOMAP, as Arnd pointed out in the previous discussion (thus, CONFIG_GENERIC_IOMAP is not really generic in this regard). So as I see it, only m68k WOULD need its own custom definition of iomem_is_ioport(). But as I understand it it doesn't because it uses the one from asm-generic/pci_iomap.h ?? I wasn't entirely sure how to deal with the address ranges for the generic implementation in asm-generic/io.h. It's marked with a TODO. Input appreciated. I removed the guard around define pci_iounmap in asm-generic/io.h. An alternative would be to have it be guarded by CONFIG_GENERIC_IOMAP and CONFIG_GENERIC_PCI_IOMAP, both. Without such a guard, there is no collision however, because generic pci_iounmap() from drivers/pci/iomap.c will only get pulled in when CONFIG_GENERIC_PCI_IOMAP is actually set. I cross-built this for a variety of architectures, including the usual suspects (s390, m68k). So far successfully. But let's see what Intel's robots say :O P. Original cover letter: Hi! So it seems that since ca. 2007 the PCI code has been scattered a bit. PCI's devres code, which is only ever used by users of the entire PCI-subsystem anyways, resides in lib/devres.c and is guarded by an ifdef PCI, just as the content of lib/pci_iomap.c is. It, thus, seems reasonable to move all of that. As I were at it, I moved as much of the devres-specific code from pci.c to devres.c, too. The only exceptions are four functions that are currently difficult to move. More information about that can be read here [1]. I noticed these scattered files while working on (new) PCI-specific devres functions. If we can get this here merged, I'll soon send another patch series that addresses some API-inconsistencies and could move the devres-part of the four remaining functions. I don't want to do that in this series as this here is only about moving code, whereas the next series would have to actually change API behavior. I successfully (cross-)built this for x86, x86_64, AARCH64 and ARM (allyesconfig). I booted a kernel with it on x86_64, with a Fedora desktop environment as payload. The OS came up fine I hope this is OK. If we can get it in, we'd soon have a very consistent PCI API again. Regards, P. Philipp Stanner (5): lib/pci_iomap.c: fix cleanup bugs in pci_iounmap() lib: move pci_iomap.c to drivers/pci/ lib: move pci-specific devres code to drivers/pci/ pci: move devres code from pci.c to devres.c lib, pci: unify generic pci_iounmap() MAINTAINERS | 1 - drivers/pci/Kconfig | 5 + drivers/pci/Makefile | 3 +- drivers/pci/devres.c | 450 +++++++++++++++++++++++++ lib/pci_iomap.c => drivers/pci/iomap.c | 49 +-- drivers/pci/pci.c | 249 -------------- drivers/pci/pci.h | 24 ++ include/asm-generic/io.h | 27 +- include/asm-generic/iomap.h | 21 ++ lib/Kconfig | 3 - lib/Makefile | 1 - lib/devres.c | 208 +----------- lib/iomap.c | 28 +- 13 files changed, 566 insertions(+), 503 deletions(-) create mode 100644 drivers/pci/devres.c rename lib/pci_iomap.c => drivers/pci/iomap.c (75%)