If a PDF is not available for free download, your paper will have a limited reach. Most publishers allow uploading a pre-publication version to an institutional repository or personal website. Use this option. Academic publishing is an outrageously lucrative business: a single article usually costs at least thirty dollars. You do not see any royalties, the editor does not get a cent, and copy-editing is outsourced for a few dollars a paper. Hence publishers register a profit margin up to 40 %. Your interest is to make your work as widely available as possible. You are legally allowed to do so, even if you signed a copyright transfer.
Before you begin
Never include your email address in a publicly available PDF. It is too easy to extract. Academic spammers with flood your inbox with calls for papers that you are not even vaguely interested in.
Using arXiv for pre-submission manuscript is a standard procedure in mathematics and physics, and many computer scientists follow suit.
This model is desirable as it reduces the time your manuscript reaches the market. Later, if the paper is accepted in a journal, it is like putting a brand on your effort. A disadvantage is that while arXiv is moderated, it does not offer peer-reviews, and occasionally rejects manuscripts without providing reasons. Contrary to expectations, the unreviewed papers have a high quality: many arXiv categories rank high as publication outlets on their own right, as measured by Google Citations.
It is uncertain whether a computer science journal welcomes papers already published on arXiv, and programme chairs at conferences are unlikely to be happy to learn that you are trying to present a paper from your arXiv account. The best way to check is to search for your target journal on arXiv. If there are papers that were later published in the journal, they will be probably fine with submitting your manuscript on arXiv.
Remember: you can never change a paper that already appeared on arXiv. If you change anything, it will appear as a new version, with the old one still available. This is one more reason never to include your email address in the paper.
If your paper is accepted in a journal or in a conference proceedings, you sign a copyright transfer, so that the publisher can shamelessly profiteer from your work while also restricting access to it. Yet, if you read the terms of the agreement carefully, most copyright transfers allow you to put your pre-publication manuscript in an institutional repository or on a personal website.
The question is which variant of your pre-publication manuscript you can distribute. Some publishers allow you putting the accepted version online. The only difference between the officially published variant will be copy-editing and formatting. Other publishers insist on the original variant, that is, the version prior to peer-review and subsequent corrections. In such a case, the version that makes it to the institutional repository will be of much lower quality.
If your institution does not have a repository, and you are associated with a European institutions, Open Access Infrastructure Research for Europe provides you a space to share your manuscript. If this is not an option, check the inclusion requirements by Google Scholar. If you put the PDF on your personal website, you want to be sure that it will be indexed the same way as it is indexed on the publisher's website. Otherwise you will end up with two versions of your manuscript on Google Scholar. Google recently enabled merging records, so this is no longer a big problem.
Few authors know that they are not obliged to transfer the copyright. It is not necessary to get your paper published in a journal or proceedings. You are allowed to retain extensive rights for distribution by attaching an author addendum to your copyright transfer. If you find the copyright transfer agreement too restrictive, do not hesitate changing the clauses.
Other information related to your manuscript
Apart from making the paper available for download, it is a good idea to gather all related information at one place, so interested readers have the fewest obstacles to learn more about your work. Related information includes the citation itself, code, slides for a conference paper, DOI link to the journal publication, links to data, and so on. Bibbase lets you manage all this by the simplest means: you only need to provide a Bibtex file of your papers. Your Bibtex entries should include the URLs for the PDF, related code, and so on. A PHP script automatically takes care of the rest, and generates a nicely formatted list of your papers.