On Thu, 22 Jul 2021 11:28:17 +0200, Dan Carpenter wrote: > > tree: https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master > head: b1347210b01daa977ea980268927aa99198ceccc > commit: 0e175f665960c7be30aba47afbee4f2d121ea5fc [1517/2389] ALSA: als4000: Allocate resources with device-managed APIs > config: x86_64-randconfig-m001-20210722 (attached as .config) > compiler: gcc-10 (Ubuntu 10.3.0-1ubuntu1~20.04) 10.3.0 > > If you fix the issue, kindly add following tag as appropriate > Reported-by: kernel test robot <lkp@xxxxxxxxx> > Reported-by: Dan Carpenter <dan.carpenter@xxxxxxxxxx> > > smatch warnings: > sound/pci/als4000.c:930 snd_card_als4000_probe() warn: 'pci' not released on lines: 852,872. > > vim +/pci +930 sound/pci/als4000.c > > e23e7a14362072 Bill Pemberton 2012-12-06 809 static int snd_card_als4000_probe(struct pci_dev *pci, > ^1da177e4c3f41 Linus Torvalds 2005-04-16 810 const struct pci_device_id *pci_id) > ^1da177e4c3f41 Linus Torvalds 2005-04-16 811 { > ^1da177e4c3f41 Linus Torvalds 2005-04-16 812 static int dev; > 17c39d9a59cbf4 Takashi Iwai 2005-11-17 813 struct snd_card *card; > 17c39d9a59cbf4 Takashi Iwai 2005-11-17 814 struct snd_card_als4000 *acard; > c0874449849175 Andreas Mohr 2008-08-20 815 unsigned long iobase; > 17c39d9a59cbf4 Takashi Iwai 2005-11-17 816 struct snd_sb *chip; > 17c39d9a59cbf4 Takashi Iwai 2005-11-17 817 struct snd_opl3 *opl3; > ^1da177e4c3f41 Linus Torvalds 2005-04-16 818 unsigned short word; > ^1da177e4c3f41 Linus Torvalds 2005-04-16 819 int err; > ^1da177e4c3f41 Linus Torvalds 2005-04-16 820 > ^1da177e4c3f41 Linus Torvalds 2005-04-16 821 if (dev >= SNDRV_CARDS) > ^1da177e4c3f41 Linus Torvalds 2005-04-16 822 return -ENODEV; > ^1da177e4c3f41 Linus Torvalds 2005-04-16 823 if (!enable[dev]) { > ^1da177e4c3f41 Linus Torvalds 2005-04-16 824 dev++; > ^1da177e4c3f41 Linus Torvalds 2005-04-16 825 return -ENOENT; > ^1da177e4c3f41 Linus Torvalds 2005-04-16 826 } > ^1da177e4c3f41 Linus Torvalds 2005-04-16 827 > ^1da177e4c3f41 Linus Torvalds 2005-04-16 828 /* enable PCI device */ > 0e175f665960c7 Takashi Iwai 2021-07-15 829 err = pcim_enable_device(pci); > 61ea06e8a88163 Takashi Iwai 2021-06-08 830 if (err < 0) > ^1da177e4c3f41 Linus Torvalds 2005-04-16 831 return err; > 61ea06e8a88163 Takashi Iwai 2021-06-08 832 > ^1da177e4c3f41 Linus Torvalds 2005-04-16 833 /* check, if we can restrict PCI DMA transfers to 24 bits */ > 669f65eaeb969e Takashi Iwai 2021-01-14 834 if (dma_set_mask_and_coherent(&pci->dev, DMA_BIT_MASK(24))) { > d85d878efb3047 Takashi Iwai 2014-02-25 835 dev_err(&pci->dev, "architecture does not support 24bit PCI busmaster DMA\n"); > ^1da177e4c3f41 Linus Torvalds 2005-04-16 836 return -ENXIO; > ^1da177e4c3f41 Linus Torvalds 2005-04-16 837 } > ^1da177e4c3f41 Linus Torvalds 2005-04-16 838 > 61ea06e8a88163 Takashi Iwai 2021-06-08 839 err = pci_request_regions(pci, "ALS4000"); > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > 0e175f665960c7 Takashi Iwai 2021-07-15 840 if (err < 0) > ^1da177e4c3f41 Linus Torvalds 2005-04-16 841 return err; > c0874449849175 Andreas Mohr 2008-08-20 842 iobase = pci_resource_start(pci, 0); > ^1da177e4c3f41 Linus Torvalds 2005-04-16 843 > ^1da177e4c3f41 Linus Torvalds 2005-04-16 844 pci_read_config_word(pci, PCI_COMMAND, &word); > ^1da177e4c3f41 Linus Torvalds 2005-04-16 845 pci_write_config_word(pci, PCI_COMMAND, word | PCI_COMMAND_IO); > ^1da177e4c3f41 Linus Torvalds 2005-04-16 846 pci_set_master(pci); > ^1da177e4c3f41 Linus Torvalds 2005-04-16 847 > 0e175f665960c7 Takashi Iwai 2021-07-15 848 err = snd_devm_card_new(&pci->dev, index[dev], id[dev], THIS_MODULE, > e58de7baf7de11 Takashi Iwai 2008-12-28 849 sizeof(*acard) /* private_data: acard */, > e58de7baf7de11 Takashi Iwai 2008-12-28 850 &card); > 0e175f665960c7 Takashi Iwai 2021-07-15 851 if (err < 0) > e58de7baf7de11 Takashi Iwai 2008-12-28 852 return err; > ^^^^^^^^^^ > Smatch complains because there is no error handling. Maybe the other > error paths use devm_ magic? Or more likely Smatch just isn't smart > enough... I believe those are false-positive. When the PCI device is already managed via pcim_enable_device(), all resources (even the ones without pcim_ or devm_) are freed automagically. thanks, Takashi
